ledger: increase locks granularity in lookupWithoutRewards

[algorandskiy]

Summary

Proposal for exclusion deltas iteration loop from trackers own lock.

1. Take a state snapshot for a single iteration (the first part of the for {} loop in lookupWithoutRewards) - dbRound, deltas, offset, rewards, inDeltas check.
2. make the nested loop over deltas lock free.

Why does it work:

say au.deltas has 10 elements that is backed by some array underneath. e make local deltas = au.deltas[0:len(au.deltas)], pointing to the same backing array. Then there are two options:

1. when receiving a new block => add element
   • if backing array has space, it is not reallocated, just appended, our deltas keeps pointing on it
   • if not enough space, then realloc for au.deltas but our deltas keeps pointing to the old backing array
2. when committing => remove elements
   • removal op is slicing, so au.deltas keeps pointing to the same backing array but to the right subset, our deltas is fine as well

The change would benefit from @icorderi work on making lru caches concurrency-safe.

Test Plan

Existing tests

[codecov]

Codecov Report

Merging #5527 (d299733) into master (f54e7a7) will increase coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #5527      +/-   ##
==========================================
+ Coverage   55.83%   55.86%   +0.02%     
==========================================
  Files         446      446              
  Lines       63258    63265       +7     
==========================================
+ Hits        35321    35342      +21     
+ Misses      25561    25559       -2     
+ Partials     2376     2364      -12     

Impacted Files	Coverage Δ
ledger/acctupdates.go	70.80% <100.00%> (-0.34%)	⬇️

... and 11 files with indirect coverage changes

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[algonautshant]

Looks good.