load balance by CUs

[apfitzge]

Problem

• Current load balancing is done by transaction count
• This leads to unbalanced loads across the threads
• Until Scheduler - prioritization fees/cost solana-labs/solana#34888 CUs were totally innaccurate. However, in recent epochs with the prioritization change, the top scheduled txs are showing significantly more accurate estimates

Summary of Changes

• Use CUs to load-balance across threads

Fixes #

[apfitzge]

Still use the number of txs as a backup comparison if the CUs happen to be exactly equal

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.1%. Comparing base (4ae2ca1) to head (f8d8109).
Report is 3 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1221   +/-   ##
=======================================
  Coverage    82.1%    82.1%           
=======================================
  Files         886      886           
  Lines      236439   236453   +14     
=======================================
+ Hits       194252   194320   +68     
+ Misses      42187    42133   -54     

[tao-stones]

lgtm

[mergify]

Backports to the beta branch are to be avoided unless absolutely necessary for fixing bugs, security issues, and perf regressions. Changes intended for backport should be structured such that a minimum effective diff can be committed separately from any refactoring, plumbing, cleanup, etc that are not strictly necessary to achieve the goal. Any of the latter should go only into master and ride the normal stabilization schedule. Exceptions include CI/metrics changes, CLI improvements and documentation updates on a case by case basis.