feature: Add dynamic resource allocation(DRA) plugin

[JesseStutler]

What type of PR is this?

/kind feature

What this PR does / why we need it:

Background

Inspired by #3798 , introduce DRA into volcano. It solves the following things:

• Introduce the related codes of The pods that successfully apply for gpu resources lack gpu card allocation information #3777, add rollback and errInfo into Pipeline and Allocate of statement.
• Refactor the backfill action to use Allocate and Unallocate in statement instead, stay consistent with other actions.
• Add a new extension point called PreBindFn, volumebinding and DRA need to use such extension point to bind resources such as ResourceClaim, PVC before binding the pod onto the node. Related issue: Introduce a new interface AddPreBindFn or AddBindFn in session #3618
• Copy the whole related codes of structured parameters DRA into volcano, because currently dynamicresources struct can't be accessed outside the k8s.io/kuberentes package, if we need to call k8s extension points such as PreFilter, Filter, etc., we have to access the struct. After this pr is merged and volcano upgrade the k8s version to 1.32: Refactor the dynamicResources struct to DynamicResources kubernetes/kubernetes#128399, we may remove these codes.
• Add a new plugin called dynamicresources, which adds some extension points such as:
  • PrePredicateFn
  • PredicateFn
  • EventHandler
  • PreBindFn

See docs/design/dynamicresources-plugin.md for more details.

Verfication

I followed up https://github.com/kubernetes-sigs/dra-example-driver to deploy an example resource driver and deploy the demo yaml gpu-test{1,2,3,4,5}.yaml, the pod can be successfully scheduled:

GPU information printed in containers:

And allocation result in ResourceClaim is successfully updated:

[volcano-sh-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by:
To complete the pull request process, please assign kevin-wangzefeng
You can assign the PR to them by writing /assign @kevin-wangzefeng in a comment when ready.

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

Needs approval from an approver in each of these files:

• OWNERS

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

[googs1025]

/cc

[googs1025]

Do we need to enable this feature-gate by default?

[JesseStutler]

What do you think? Do we need to enable it by default? I think that if users need to enable dra plugin, they also need to reconfigure the scheduler plugins configmap to enable it, it also looks like featuregate. If we don't enable this feture-gate, users need to do reconfiguring the scheduler plugins configmap and enabling the feature-gate two steps.

[JesseStutler]

OK we need not to enable this feature-gate by default, I will add some informers in schedulerCache, some users may need to use k8s below v1.31, if there are no APIs like DeviceClass, ResourceClaim, etc registered, it will cause some errors.

[googs1025]

Is this part directly imported into the scheduler code? Or has it been modified and adapted (I don't know much about it)? If not, should it be marked in the code?

[JesseStutler]

I just changed the New to NewDRAPlugin and deleted EventsToRegister and some related codes with it, I think volcano doesn't need them right now. These codes are copied from the kubernetes repo, so the license needs to keep.

[googs1025]

Understand, if so, if the upstream code changes, do I need to make the same changes?

[JesseStutler]

Yes, before the dynamicResources struct changed to DynamicResources, we can only copy the whole codes related into volcano, otherwise we can't directly call those extension points. It may cause us to miss tracking some repaired codes, but if we can access the DynamicResources struct in v1.32, the problem will be solved.