fix: honor kbagent action retry policy

[weicao]

Problem

Redis reconfigure validation found that lifecycle actions needing bounded wait can be killed before convergence even when the addon declares a larger timeout:

• timeoutSeconds: 120 is capped by kbagent at 60 seconds.
• retryPolicy declared on the action is registered into the kbagent action table, but the action service only used req.RetryPolicy when executing the action.
• The reconfigure path currently passes runtime ReconfigureArgs, but not a request-level retry policy, so an action-level retryPolicy is ineffective on this path.

This makes the addon-side pattern exit non-zero + retryPolicy unreliable for reconfigure actions.

Solution

• Raise kbagent maxActionCallTimeout from 60s to 120s and update the public Action timeout docs.
• Resolve retry policy the same way timeout is resolved: request-level retry policy overrides action-level retry policy; otherwise the action-level policy is used.
• Apply retry handling to non-blocking calls without runtime arguments as well.

Validation

Local validation passed:

go test ./pkg/kbagent/service -count=1
go test ./pkg/kbagent/... -count=1
make test-go-generate
go test ./pkg/kbagent/service ./pkg/controller/instanceset ./pkg/controller/instance -count=1
git diff --check

Additional test note:

• go test ./pkg/controller/component -count=1 could not run on this machine because envtest assets are missing: /usr/local/kubebuilder/bin/etcd not found. The package did not reach product assertions.

Boundary

This PR is a controller/kbagent candidate for the Redis task #46 timeout/retry gap. It still needs exact-image runtime validation on the affected reconfigure scenario before being treated as test-accepted or ready for non-draft review.

[apecloud-bot]

Auto Cherry-pick Instructions

Usage:
  - /nopick: Not auto cherry-pick when PR merged.
  - /pick: release-x.x [release-x.x]: Auto cherry-pick to the specified branch when PR merged.

Example:
  - /nopick
  - /pick release-1.1

CLA Recheck Instructions

Usage:
  - /recheck-cla: Trigger a re-check of CLA status for this pull request.
Example:
  - /recheck-cla