A form of toot search has been requested since April 2017, multiple times:
I have announced that am going to implement it with constraints that won't let it be used for abuse on Dec 20 2017, then again on Dec 31:
During implementation I seeked feedback #onhere, submitted PR on Feb 3 2018:
Then announced it in a progress report on Feb 21:
After the progress report, the first release candidate came out on Mar 2, and was available for testing and feedback for 8 days until yesterday.
The biggest issue with ElasticSearch (that was the only issue raised during all that time) is that it takes a lot of RAM and it's an extra dependency. So it was made completely optional, not everyone wants to deal with that.
It (quite obviously) only operates on data that was in the PostgreSQL database anyway. It took 78 lines of code (mostly one file) to add the indexing (which is part of the reason it was chosen - ease of integration).