[performance] Database optimizations

[tsmethurst]

This PR includes a database migration to add indexes on commonly-selected fields in database entries.

In testing on https://goblin.technology, this saw some queries go from taking about 150ms to taking 0.5ms, a 300x improvement.

From Matrix chat:

the most expensive queries (in the order of 100+ms) tend to be counts of statuses and complicated joins

in order to serialize a notification in the manner that masto expects, GtS has to do a lot of counting from the database (count number of statuses posted by this account, count number of replies/likes/faves on this status, that kind of thing)

so it runs queries like the following

SELECT "status"."created_at" FROM "statuses" AS "status" WHERE (account_id = '01Q0ER4N4DYAPM3141BZQ7B985') ORDER BY "id" DESC LIMIT 1

^^ so that finds the time of the most recent status by account 01Q0ER4N4DYAPM3141BZQ7B985

but statuses so far aren't indexed on anything except their primary key

so i believe adding indexes for things like account_id, uri, etc, should speed the selects and counts up, albeit at the cost of slower insertion times (which isn't really an issue anyway)