Improved deduplication logic to prevent skipped files during import
Some of these bugs are duplicates, but the core issues where:
Considering a TRACK_ID (
Track.mbidin Funkwhale) was unique, for instance leading to orphaned/skipped uploads depending on the processing order. Solved by doing the unicity check with the
Not including the disc number when trying to deduplicate tracks, leading to skipped uploads when a track with the same title was present at the same position but in a separate disc
I'm still testing this locally with real-world data, hence the wip.