[release-1.29] Add flag to enable fast delete of failed VMSS

[willie-yao]

What type of PR is this?

/kind cleanup

What this PR does / why we need it:

"Refactor" as a part of fork-upstream realignment. Fork introduces an environment variable AZURE_ENABLE_FAST_DELETE_ON_FAILED_PROVISIONING to enable fast delete of failed VMSS, whereas upstream does this by default.

Which issue(s) this PR fixes:

Fixes #

Special notes for your reviewer:

Does this PR introduce a user-facing change?

Question: should this be enabled by default to preserve current behavior upstream?

Add flag to enable fast delete of failed VMSS

[willie-yao]

/assign @comtalyst

[jackfrancis]

@willie-yao is there a reason we're merging this directly into the 1.29 release branch as opposed to landing in main and then backporting to the desired release branches?

[willie-yao]

@jackfrancis The fast delete that exists in upstream is only on 1.28+ and is also handled differently across versions. I also observed other deforking PRs merging directly into a release branch.

[jackfrancis]

@willie-yao makes sense

Are we able to classify this change in a semver-agreeable manner that would be appropriate for a future patch release? (i.e., bug)

[willie-yao]

I wouldn't call this a bug and I see other defork-related PRs just titled as "chore". Would that be sufficient? Not sure if defork cleanup can be a special case as a patch release item, but I just wanna keep things consistent.

[comtalyst]

/lgtm
/approve
/hold

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: comtalyst, rakechill, willie-yao

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:

• cluster-autoscaler/cloudprovider/azure/OWNERS [comtalyst]

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

[comtalyst]

It is something in between a bug and an improvement.
The nature of the issue is that some failed VMs won't get deleted fast enough. It could be argued that it is supposed to. Thus, this flag should contain an experimental method to deal with that.

[comtalyst]

/lgtm

[k8s-ci-robot]

New changes are detected. LGTM label has been removed.