Note this is actually how the postgres sql stemmer / tokenizer works, we have some workarounds in place for edge cases like urls where it can get confusing, but overall we outsource a lot of this stuff to pg.
Interestingly we did have a hack in place a few years back for “extra indexing in URLs” that @tgxworld removed due to index bloat.
I guess what I can say is that yes, we are thinking about these kind of edge cases in search, it takes quite a lot though for us to push towards hacking around the existing framework in pg fulltext.