create makedb function

[thatbudakguy]

it will considerably speed up future work to be able to make and export phonetic n-gram indices that can be later used for querying, in the same way that you can do makeblastdb and then later query using this "database". there are lots of options for export/import formats, including:

• something involving spaCy's own Lookups.from_disk()
• the LMDB "lightning" database, which BLAST uses
• something involving pickle or a python-native solution

relatedly, this command (makedb) could also be parameterized by a stored spaCy model, using Language.from_disk(). that way we could share a compiled version of the "DIRECT model" that includes our current spaCy stack.

[thatbudakguy]

See also Qdrant, a vector database that supports running in-memory, on disk, or via a dedicated server.