I'm guessing the smartest usage is to buy spells in the specialties of the covenant Magi, that are just a bit above their capabilities, so that when they get to that point, they can just spend a single season upgrading the maximum power they can put out? Or is there a better way to go about it? Maybe a bunch of low level utility spells so that the Magi aren't spending their first spell levels (the point when it's easiest to learn spells) on utilities, since they can learn a bunch of them per season? Or a mixture of the above? What do you guys think? Obviously, such perfect accommodations are a bit immersion-breaking in actual play, but I'd still like to know, if only for educational reasons.
Spells they can actually learn in the near future is a good start. It's utterly pointless to have level 50 spells in your library if no-one can learn them even from the text, unless you're trying to set up your covenant as a place where magi come to learn rare and powerful spells.
The very low end spells are less efficient, unless you plan on learning a bunch of them at once (InVi is good for this). Learning a lot of low level spells in different arts is best done through a teacher if you can find an accommodating magus. Spell instruction trading can save both magi several seasons of time.
Formulaic spells are a great way to increase the power of most magi, as they represent things they can do again and again while making full use of their magical power. Also, they can be done without the risk of a botch.
Lab texts allow you to save a lot of time while inventing spells. So IMHO you want three kinds of spell lab texts in your library.
First type of those utility spells that you will cast often, such as those InVi spells or those flexible spells that you can put to many uses (things like Unseen Porter). Many of these are relatively low level, but they may be outside of your magus' specialty. You want many of these.
Second type are low-level combat spells. To magi for whom they are out of their specialty, they give a backup attack or defense spell for those times when their specialty cannot be used (using a Pilum of Fire against a fire elemental is rather pointless). Even to the specialist, they provide a nice punch because of the higher Penetration that is possible with those spells. You want a small number of these.
For these two types, you want to group those spells into a limited number of Technique+Form combinaisons, so that you can maximize the lab total of various magi by allowed more than one spell to be learned at once. The specialist may be able to learn 3-4 of them in a single season, while even the magus who has no particular skill should still be able to learn them one at a time.
The third type is those spells that the specialist may be able to invent, but that would take multiple seasons. If the specialist's lab total is 40, then inventing a level 30 spell is possible in 3 seasons. But with the lab text he can do it in a single season. That is even more true when the lab total is just a few points over of the spell's level (lab total 41 vs spell level 40).
I know this might be taking the conversation in another direction, but it might be reasonable to conclude that Lab Texts are fantastically expensive based on build points.
A level 40 spell costs 8 build points. That's the same as 40 pawns of vis as vis stocks (or you can go the other way and vis stocks are fantastically cheap). Could those vis stocks be used in trade to acquire lab texts for needed spells? I suppose one way that the high price can be justified is if the PCs are designing the library of texts for their own specialties, so that they mesh perfectly. But if the players are generating a lab for verisimilitude then I would focus on granting more free build points for certain important text such as Aegis of a couple of different levels, Wizard's/Ritual Communion of several levels and then several other common spells from Romer's Book[1].
If you're designing the lab texts as the SG, I'd probably just put what I want and charge some fraction of build points to account for it. And then you can tweak it to have a few spells they can learn early and a few spells they can reach to acquire in a few years.
[1] I'm a player in that saga mentioned in the linked post, and there's a house rule reducing the cost of lab texts. Although this rule is for use in play and not necessarily covenant design using build points, I think it's a good rule of thumb to use when determining what kind of library of lab texts a covenant that exists should have...
You speak only the truth. Especially the low level, common spells (5-15). To put it in perspective, a professional scribe (scribe 6) trained in Magic Theory can copy 360 levels of spell texts in a season. That same scribe could copy a single level 12 summae in the same season of work. It seems logical that essentially the same amount of raw materials is going into each - i.e. lab texts are far less weighty than summae, in general. So the physical cost of the two should be little different. But a L12 Q12 Summae is 24BP, while 360 levels of lab texts are 72 BP. Granted it took a magus to clarify the lab text at some point, but in most cases scribes will be working from texts that have been clear for decades, possibly centuries.
I agree with John Prins' point (agreeing with Jonathan Link). And, to add to it, at some point a magus must have written the tractatus/summa to be copied just as a magus must have translated the laboratory notes. So that shouldn't be a big mark against laboratory notes at all.
I agree 100%.
However - the premise of the OP worries me from a "let's at least pretend to be realistic" perspective (and he acknowledges this - but read on). Few libraries are set up to be "efficient", unless the library is specifically designed to be for learning and teaching*, and even then the students themselves are largely unpredictable unless they are fledgeling apprentices with 0's in all their Arts.
(* Most libraries are used for that, yes - but few would be designed for it. Most are repositories of ad hoc collections of material, accidental and almost random. Some collections have a pattern to them, but rarely is that pattern "to teach these spells to the magi who are here now".)
More, one might feel forced to ask - does the concept of "efficient" presume fore-knowledge of the Te/Fo Lab Totals of those who will be learning the spells? That is, do we know that they have a Lab Total of 25, etc? And if so... how was that known when the Library was set up? 
("Believability" does not have to be a slave to "realism" - it's a game, not a simulation - but otoh, "totally unbelievable and inexplicable" is rarely acceptable. I'll pass on "bibliotheca ex machina" every time. ysmv.)
When I want to give the Players "their build-points worth" for a library, I'll try to combine efficiency with a dash of realism, to include "collections" that have a useful (and coincidentally magnitude-appropriate) theme to them. Perhaps a Jerbiton or Gifted Redcap made a collection of spells that helped traveling - keeping warm, keeping dry, wards against unpleasant things for the night, spells for dealing with uncooperative innkeepers or suspicious guards, things like that. Or a Flambeau compiled their favorite (low-medium level) combat spells all into one volume - how convenient! Or (best?), a previous mage of THIS covenant made sure to collect spells that helped with known problems in the immediate environment - those faeries in the lake, and water breathing, and seeing underwater, and keeping boats from sinking, and Scrying Pool, and so forth.
Collections like these can be tweaked (a spectrum of magnitudes and Arts) so that almost anyone can get a full season's worth of study out of them, AND benefit from spells they almost certainly would not have taken out of Gauntlet, but are highly useful "here, now". That combines plausibility with "efficiency", win/win. 
Add in some predictably useful generic spells - Invis, CrCo healing, Inexorable Search*, some TeVim effects (Aegis, Wizard's communion, Collect Essence of Beast, some PeVi and InVi, etc) - and some truly random(?) bits for color and fluff (that may prove useful waaaay down the road, in some future plot arc?), and you're set.
(* Don't give away anything that will make your job too hard - if no one has InMe effects, you don't have to give them away!) 
As a side note, literal "accommodations" are not unrealistic nor very difficult to make. If you have the Covenants supplement, the section on Labs describes how specialized labs for Learning/Inventing are actually pretty easy to setup and can be cheap to maintain, and do not have to take up a lot of room.
(Going from memory here, but there are about a half-dozen "activities" that a lab can be specialized in, and one of the major flaws/virtues is to limit/focus a lab toward just one of those. This allows it to be very small, very cheap, and add a nice bonus to the activity. One for learning/inventing new spells, one for copying, one for extracting Vis - whatever suits the Covenant's fancy.)
I find it pretty common to have the "copying" lab to be synonomous with the scriptorium. Ie, it doesn't need to be a "lab" in the classic "alchemy and alter" sense. The season of setup time to get the scripting bonuses can simply be re-arranging the feng shui of the scriptorium according to Hermetic principles. From what I recall, anyone in the "lab" can gain the bonuses - meaning that you can hire a couple of scribes, and add +3 (or whatever the bonus is) to their Scribe ability.