This is just my guess on this, I've never seen this discussion arise in the mailing lists (that doesn't necesarily mean it didn't take place, though).
But I don't think there's any technical reason why this couldn't happen, other than the lack of interest and/or man power. I can think at least 5 ways to achieve this, and that only in a quick glance, there are surely many more,
- profiles, those can have custom masks, uses, keywords...
- keywords, for example, we could add a lot of greys in between the white (let's say "x86") and the black (~x86)
- overlays, these can implement a combination of the two above
- alternative/decoupled portage trees, as you suggested. IMHO not necessary, since it would take just some kb of metadata to difference an LTS tree from a standard one
Those are not necessarily exlusive at all.
The problem is that Gentoo has always been understaffed. So (again my personal guess, which might be completely incorrect) the problem could be that. Plus the fact that the do-it-y/self phylosophy that reigns soverein between the Gentoo users doesn't make this a top-priority.
If you truly want to discuss this and iron it a bit, you might be interested in joining the gentoo-user mailing list, and then, once the idea matures, join some other more specific list such as gentoo-server, gentoo-security and/or gentoo-releng. For this to happen, a proper Gentoo Project
, backed by a team of interested persons should be founded.