Please no offense, but I don't give a rat what your background is- is that relevant? I'd consider if so! This feels like an interrogration-first line of questioning, and I'm trying to help us work out what's going on, and help us find relevant facts to begin to justify your line of inquiry.
How many kids learning to program on the X or Y website do you expect will miss the 0.3ms extra it took for the site's offline capable web-ide to work? What harm, what badness is there to doing this on the web, and why would you call out a JS implementation up front as negative?
That's really it- I still don't see you having any legitimate grievance besides whatever insane use case you have in mind for yourself and yourself alone needing 4GB crazy town git repos (you, linux, and chrome may indeed have problems). I just don't think that kind of extreme performance concern is where every single project ought start- some other well known programmers and I know it's better to not worry too greatly about performance until you have the problem in the first place. Focus on what can be done first, what good can be created, and then refine that good; don't nip good in the bud because down the road it won't do Y: that's premature man, and not constructive. Can we agree that JS projects like Three.js provide a lot of value, have allowed a lot of cool things to get built that people might not otherwise have made, that at least as importantly many of people might not have seen? Are they slower? Maybe, yeah, probably. My suggested reading here is Steve Wittens, writing about WebGL- there's a niche served, and allow that niche to enrich itself was more or less the take away for me, even though dieties and demigods like @id_aa_carmack might not have immediate use- http://acko.net/blog/on-webgl/
Kinda felt like you ran in here and farted on our rainbows, to twist around your metaphor. The only concern you've stated is performance- let's pretend for a second that it's ok that you're calling it into question before we have any guestimates much less estimates for what performance is going to be- what's your estimate? For 98% of the projects on Git, do you think the performance of a JS git impl is going to be sufficiently fast as to not be problematic? Will it even be a noticeable factor, the speed, compared to cmd line git tools? If those users are happy and feel served, is that ok? Do you have concerns other than performance that I ought take into consideration when considering your expressed concern about this being a JavaScript library?