Optimize Counter thread-to-stripe distribution

[puzpuzpuz]

• The number of stripes is now calculated dynamically based on the number of cores and GOMAXPROCS
• The performance is close enough to the old one with 64 stripes while the memory footprint is significantly lower and, what's also important, unlucky distribution isn't a problem anymore

Note: this is a breaking change due to the new NewCounter method.

[codecov-commenter]

Codecov Report

Base: 97.03% // Head: 96.35% // Decreases project coverage by -0.68% ⚠️

Coverage data is based on head (9d52599) compared to base (051117f).
Patch coverage: 91.17% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #71      +/-   ##
==========================================
- Coverage   97.03%   96.35%   -0.69%     
==========================================
  Files           6        6              
  Lines         912      932      +20     
==========================================
+ Hits          885      898      +13     
- Misses         22       27       +5     
- Partials        5        7       +2     

Impacted Files	Coverage Δ
util.go	85.71% <81.25%> (-14.29%)	⬇️
counter.go	100.00% <100.00%> (ø)
rbmutex.go	87.17% <100.00%> (-7.70%)	⬇️
map.go	96.78% <0.00%> (-0.23%)	⬇️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.