Fixed Style hash

[willmcgugan]

Fixes a performance bug in the Style class.

Essentially the hash calculation was inconsistent. Things that were equal would sometimes have different hashes. I noticed this when adding lru_cache to Style.__add__. What should have been an easy win caused dramatically reduced performance, because it broken the mechanics of storing Styles in a dict.

The hashing is now in once place. It is also lazy, so the hash is stored the first time it is used.

[codecov-commenter]

Codecov Report

Merging #2346 (3bc9b28) into master (14d47c9) will decrease coverage by 0.17%.
The diff coverage is 96.61%.

@@            Coverage Diff             @@
##           master    #2346      +/-   ##
==========================================
- Coverage   98.88%   98.71%   -0.18%     
==========================================
  Files          73       73              
  Lines        7629     7761     +132     
==========================================
+ Hits         7544     7661     +117     
- Misses         85      100      +15     

Flag	Coverage Δ
unittests	98.71% <96.61%> (-0.18%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
rich/_export_format.py	100.00% <ø> (ø)
rich/live.py	97.90% <40.00%> (-2.10%)	⬇️
rich/_win32_console.py	89.94% <92.30%> (-1.54%)	⬇️
rich/console.py	98.29% <96.03%> (-0.49%)	⬇️
rich/style.py	99.76% <96.15%> (-0.24%)	⬇️
rich/_inspect.py	100.00% <100.00%> (ø)
rich/_lru_cache.py	100.00% <100.00%> (ø)
rich/_wrap.py	100.00% <100.00%> (ø)
rich/box.py	100.00% <100.00%> (ø)
rich/cells.py	100.00% <100.00%> (ø)
... and 9 more

---

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 761cc8d...3bc9b28. Read the comment docs.

[olivierphi]

LGTM 🙂