add restore finalizer to clean up external resources

[allenxu404]

Thank you for contributing to Velero!

Please add a summary of your change

• Add restore finalizer to clean up external resources when restore is being deleted

Does your change fix a particular issue?

Fixes #2697

Please indicate you've done the following:

• Accepted the DCO. Commits without the DCO will delay acceptance.
• Created a changelog file or added /kind changelog-not-required as a comment on this pull request.
• Updated the corresponding documentation in site/content/docs/main.

[codecov-commenter]

Codecov Report

Merging #6479 (1fbc1ac) into main (0945879) will decrease coverage by 0.27%.
The diff coverage is 53.42%.

@@            Coverage Diff             @@
##             main    #6479      +/-   ##
==========================================
- Coverage   60.31%   60.04%   -0.27%     
==========================================
  Files         229      229              
  Lines       24319    24552     +233     
==========================================
+ Hits        14667    14742      +75     
- Misses       8647     8791     +144     
- Partials     1005     1019      +14     

Impacted Files	Coverage Δ
pkg/controller/restore_controller.go	64.53% <53.42%> (-0.31%)	⬇️

... and 4 files with indirect coverage changes

[ywk253100]

@allenxu404 Does this PR cover the upgrade case? What will happen to the existing Restores?

[allenxu404]

@allenxu404 Does this PR cover the upgrade case? What will happen to the existing Restores?

@ywk253100 "The existing restores will not have a finalizer added and they will be deleted directly the way it was.

[ywk253100]

@allenxu404 Current changes that filter based on different events are complicated, how about that we handle the restore directly without considering the event type as follows:

1. Get the restore
2. If the restore doesn't exist, return directly
3. If the deletion timestamp isn't empty, clean up the resource in store, remove the finalizer, and return
4. Reconcile the finalizer if the restore isn't set. - This covers the upgrade case
5. Process the restore if it is new as previous

An example is https://github.com/kubernetes-sigs/cluster-api/blob/main/internal/controllers/machine/machine_controller.go

[ywk253100]

And please also the new code is covered by the unit test

[allenxu404]

@allenxu404 Current changes that filter based on different events are complicated, how about that we handle the restore directly without considering the event type as follows:

1. Get the restore

2. If the restore doesn't exist, return directly

3. If the deletion timestamp isn't empty, clean up the resource in store, remove the finalizer, and return

4. Reconcile the finalizer if the restore isn't set. - This covers the upgrade case

5. Process the restore if it is new as previous

An example is https://github.com/kubernetes-sigs/cluster-api/blob/main/internal/controllers/machine/machine_controller.go

@ywk253100 Yeah, make sense. I made an enhancement to the code and it functions following the procedures outlined aboved.

[ywk253100]

Let's replace resource with finalizer attached with restores as the finalizer is the internal mechanism

[allenxu404]

if we output "Waiting for restore to be deleted" to the users, users might be confused about the need for an additional operation to delete the restores. What's your thought?