Fix panic when process RunScorePlugins for cap out of range

[kerthcet]

What type of PR is this?

/kind bug
/sig scheduling
/kind regression

What this PR does / why we need it:

The root cause is we can randomly customize the preScore & Score plugins in the KubeSchedulerConfiguration so there maybe 0 score plugins but x prescore plugins, then as long as one prescore plugins returns skip, we'll panic as the cap is negative, reporting panic: runtime error: makeslice: cap out of range

numPlugins := len(f.scorePlugins) - state.SkipScorePlugins.Len()

The precise number of numPlugins should be plugin in f.scorePlugins also in SkipScorePlugins, then we will have to traverse these two slices for comparison, considering the plugin number will not large, so I pick the len(f.scorePlugins) instead.

This bug was introduced in #115652, so it's a regression.

Which issue(s) this PR fixes:

Fixes #121630

Special notes for your reviewer:

Does this PR introduce a user-facing change?

Fixed a regression since 1.27.0 in scheduler framework when running score plugins. 
The `skippedScorePlugins` number might be greater than `enabledScorePlugins`, 
so when initializing a slice the cap(len(enabledScorePlugins) - len(skippedScorePlugins)) is negative, 
which is not allowed. 

Additional documentation e.g., KEPs (Kubernetes Enhancement Proposals), usage docs, etc.:

[k8s-ci-robot]

This issue is currently awaiting triage.

If a SIG or subproject determines this is a relevant issue, they will accept it by applying the triage/accepted label and provide further guidance.

The triage/accepted label can be added by org members by writing /triage accepted in a comment.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: kerthcet

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• pkg/scheduler/OWNERS [kerthcet]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[kerthcet]

Panic here when revert the change.

[kerthcet]

/hold
Let me check again.

[kerthcet]

/hold cancel
functionality is ok.

[denkensk]

/lgtm

Thanks @kerthcet

[k8s-ci-robot]

LGTM label has been added.

Git tree hash: 588774c1c03e1f39fc06eb24c15f14f0847121e4

[alculquicondor]

Does this affect default scheduler? Do we need to cherry-pick?

[alculquicondor]

Please add a release note.

[Huang-Wei]

If it's a regression, let's ensure we tag it as kind/regression and also point out the PR that introduced this. @liggitt has a tool analyzing all regressions based on this tag.

[kerthcet]

Does this affect default scheduler? Do we need to cherry-pick?

Yes, I'll cherry-pick this.

If it's a regression, let's ensure we tag it as kind/regression and also point out the PR that introduced this

Done.

/kind regression

[liggitt]

Thanks, please add the specific release (1.x.y) that regressed in the release note

[kerthcet]

Thanks, please add the specific release (1.x.y) that regressed in the release note

Do I have to add the patch version? This bug was introduced since 1.27, so basically 1.27.X and 1.28.X should all have this bug. Or what I need to add is simply 1.27, 1.28? @liggitt

EDIT:
Modified as Fixed a regression since 1.27.0 ... plz ping me if not meet your expectation.

[alculquicondor]

Just to clarify, this affects the default scheduler, but not the default profile, as we never have a Score plugin enabled without its PreScore counterpart.
Is that correct?

[kerthcet]

Yes, for default profile, preScore & score are paired, but for default scheduler, we can customize another profile which breaks the pairing. I used to suggest validating the pairing, they should be enabled and disabled together, then we can also remove the failover like preScore state not exist, but seems not accepted at that moment, I forgot the reason. :(

[Huang-Wei]

The upgrade of scheduler-plugins also hit this regression: kubernetes-sigs/scheduler-plugins#670 (comment)