The real reason is mostly just speed. Pleroma has a full text search and its frankly kinda ass in terms of speed. You need fairly beefy hardware to make full-text searches an option if you're dealing with that many "documents". Doubly so on Mastodon, which is already fairly bogged down by being a Rails application.
The infra can't really support it without heavily centralizing the model (something for which Mastodons questionable tech stack already doesn't do itself any favors).
There's also a bit of a privacy concern in that many instances just don't like it when big nameless entities start scraping the posts of their users, which also leads to a fairly hostile mentality towards the concept.