fix: evaluate after filter

[gaius-qi]

Since the final length of the filter is the candidateParentLimit used, the parents after the filter is the returned parents.

Description

Related Issue

Motivation and Context

Screenshots (if appropriate)

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation Update (if none of the other choices apply)

Checklist

• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have read the CONTRIBUTING document.
• I have added tests to cover my changes.

[codecov]

Codecov Report

Merging #2363 (cc206ea) into main (39c9839) will increase coverage by 0.00%.
The diff coverage is 66.66%.

@@           Coverage Diff           @@
##             main    #2363   +/-   ##
=======================================
  Coverage   49.70%   49.70%           
=======================================
  Files         148      148           
  Lines       19644    19646    +2     
=======================================
+ Hits         9765     9766    +1     
  Misses       9267     9267           
- Partials      612      613    +1     

Flag	Coverage Δ
unittests	49.70% <66.66%> (+<0.01%)	⬆️

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

Impacted Files	Coverage Δ
scheduler/scheduling/scheduling.go	84.60% <66.66%> (+0.36%)	⬆️

... and 1 file with indirect coverage changes

[bigerous]

I suggest evaluating and sorting candidates peers before filtering them for these reasons:

• the complexity of evaluating is far low than filtering. The filtering need Lock on DAG, may slow the speed of scheduling. The filter process will stop at once when there is enough candidates(it would be 4 candidates if filtering after sort, but 40 when sort after filter)
• The scheduling is not atomic(sometimes a candidate peer is good when filtering, but may become bad because assignments on other scheduling goroutines), so the process is the faster the better
• when evaluatoring and sorting, it would be better if more peers are included.

[hyy0322]

lgtm