Cache: update add_chunk to use ChunkIndex.add

[enkore]

Also fixes a small cosmetic bug (would only affect recreate) where with add_chunk(overwrite=True) chunks that aren't unique would count towards the unique/deduplicated size.

[codecov-io]

Current coverage is 84.59%

Merging #925 into master will not affect coverage as of 5b36651

@@            master    #925   diff @@
======================================
  Files           14      14       
  Stmts         4778    4778       
  Branches       868     868       
  Methods          0       0       
======================================
  Hit           4042    4042       
  Partial        213     213       
  Missed         523     523       

Review entire Coverage Diff as of 5b36651

---

Uncovered Suggestions

1. +0.32% via borg/helpers.py#897...911
2. +0.25% via borg/upgrader.py#316...327
3. +0.23% via borg/archiver.py#806...816
4. See 7 more...

Powered by Codecov. Updated on successful CI builds.

[ThomasWaldmann]

did you check all callers of ChunkIndex.add whether they want / don't want to update size/csize?

[enkore]

The only direct user of this is the Cache. In add_chunk that's what we want, in the index building it's not relevant: In which order the archive indices are merged is dictionary order thus random, hence, if there was any inconsistent csize data in the archives, the effect is random, independent of which side merge/add take the csize from.

size is only different for either corruption / logic bugs or hash collisions.