Use 'parallel' policy for workspace pod rollouts to avoid stalls

[EronWright]

Proposed changes

This PR seeks to address this issue (k8s: "Forced rollback") that occurs when the workspace pod is in a crashloop:

When using Rolling Updates with the default Pod Management Policy (OrderedReady), it's possible to get into a broken state that requires manual intervention to repair.

The parallel policy seems to enable the statefulset controller to forcibly remove a pod when a new revision is available. The controller seems to obey the termination grace period as is important, and I can't think of any other negatives. But there's a concern in the k8s community about this approach: kubernetes/website#47085

Note that a workspace consists of one replica, and is rather like a singleton with good behavior w.r.t. Pulumi state locking and compatible with persistent volumes.

Related issues (optional)

Closes #801

[codecov]

Codecov Report

Attention: Patch coverage is 12.50000% with 7 lines in your changes missing coverage. Please review.

Project coverage is 51.16%. Comparing base (fc48798) to head (4e33702).
Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
...r/internal/controller/auto/workspace_controller.go	12.50%	7 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #802      +/-   ##
==========================================
- Coverage   51.22%   51.16%   -0.06%     
==========================================
  Files          31       31              
  Lines        4318     4325       +7     
==========================================
+ Hits         2212     2213       +1     
- Misses       1917     1923       +6     
  Partials      189      189              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rquitales]

Per offline discussion, the workaround makes sense, but we should somehow exercise this in an e2e test to ensure that taking advantage of the implementation detail of parallel for StatefulSet updates carries through for future Kubernetes releases.

Ref from docs:

This option only affects the behaviour for scaling operations. Updates are not affected.