fix(binder): preserve reservations during bind cache lag - #2049

Description

Fixes reservation pod cleanup during BindRequest/pod cache lag.

When SyncForGpuGroup sees a reservation pod but does not see any fraction pods for the GPU group, it uses BindRequests as a guard before deleting the reservation. The previous guard ignored succeeded BindRequests. In practice, the BindRequest status can advance before the pod GPU-group label is visible through the client cache, so cleanup could delete the reservation pod in the middle of a successful bind.

This keeps succeeded BindRequests protective while their target pod is still live. Failed BindRequests and succeeded BindRequests whose pod is gone or terminal do not protect the reservation.

Related Issues

N/A

Checklist

Note: Ensure your PR title follows the Conventional Commits format (e.g., feat(scheduler): add new feature)

• Self-reviewed
• Added/updated tests (if needed)
• Updated documentation (if needed)
• Added a changelog fragment via make changelog (or applied the skip-changelog label). Do not edit CHANGELOG.md directly — pending fragments are folded into it at release time.

Breaking Changes

None.

Additional Notes

Validation run:

go test ./pkg/binder/binding/resourcereservation -count=1
KUBEBUILDER_ASSETS="$(pwd)/$(bin/setup-envtest use 1.34.0 -p path --bin-dir bin)" go test ./pkg/binder/controllers/integration_tests -run TestAPIs -count=1
git diff --check