Patch aggregate statistics

[ThomasTNO]

I discovered that the aggregated statistics were not properly computed.

In particular in case of file movements, file deletions and deeply nested directories.

This PR should resolve the issue. Please validate if this does not break other functionality (as I am relatively new to the code base.)

[codecov-commenter]

Codecov Report

Merging #139 (f772c8a) into master (e72b7d9) will decrease coverage by 0.78%.
The diff coverage is 83.72%.

❗ Current head f772c8a differs from pull request most recent head d8e2668. Consider uploading reports for the commit d8e2668 to get more accurate results

@@            Coverage Diff             @@
##           master     #139      +/-   ##
==========================================
- Coverage   97.82%   97.04%   -0.79%     
==========================================
  Files          23       23              
  Lines        1152     1183      +31     
==========================================
+ Hits         1127     1148      +21     
- Misses         25       35      +10     

Impacted Files	Coverage Δ
src/wily/archivers/filesystem.py	100.00% <ø> (ø)
src/wily/state.py	98.80% <ø> (ø)
src/wily/archivers/git.py	91.66% <81.25%> (-8.34%)	⬇️
src/wily/commands/build.py	94.50% <83.33%> (-2.37%)	⬇️
src/wily/archivers/__init__.py	100.00% <100.00%> (ø)
src/wily/cache.py	98.18% <0.00%> (-1.82%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e72b7d9...d8e2668. Read the comment docs.

[ThomasTNO]

I fixed most unit tests, two failing tests remaining. Feel free to suggest a solution to those.

[tonybaloney]

Thanks for this! Since this changes the wily cache spec, have you tested what happens when you have a cache built with an older version?

[ThomasTNO]

Thanks for this! Since this changes the wily cache spec, have you tested what happens when you have a cache built with an older version?

I did not test this. I guess that fails, as the Revision() variables changed. Anyhow, my suggestion would be to rebuild the cache as the previous aggregates were faulty.

[ThomasTNO]

I just ran it on cache built with an older version, it seems to just work fine.