Master/global file search index/database
Searching for filenames/hashes is faster than in v3.18 and earlier, but it could be faster:
Generate/use a common search database (ala locate/mlocate/plocate for *nix). A simple approach would be to generate or update a master index which includes all fileidxrec_t records (sorted by name) and the internal code of the relevant directory. Such a file could get pretty big, but still searching one big file would be noticeably faster than searching through many smaller .SID (Synchronet Index) files.