Hydrusvideodeduplicator

Bugs Fixed:
- Database batches were always 1 for clearing trashed files and updating the search cache
- Potentially a minor performance improvement for large databases

Features:
- Add parallelized hashing
- You can set the number of CPUs with `--job-count`. Default is all CPUs but one (-2).
- This may use a lot of memory if you're hashing large video files. The entire video is loaded into memory before hashing. If performance degrades then try lowering the number of jobs.

Bug Fixes:
- Fix search cache bug
- Fix bad build config from 0.2.4

Search cache was becoming invalid after deleted files were removed from the database, so some files were being skipped on the subsequent runs.

This should have only affected users who deleted a ton of files and ran the program multiple times.
You should probably run `--clear-search-cache` on your next startup to make sure all files are compared against each other.