resourceclaim controller: deleted claims stay visible in the MutationCache for up to 1h

Finding: In the Kubernetes resourceclaim controller, a reporter states that a deleted generated ResourceClaim can remain visible through the controller's claim MutationCache for up to 1 hour, delaying recreation. This is an external report, not reproduced by KFA. Source: https://github.com/kubernetes/kubernetes/issues/142870 Reported conditions: The issue says the claim MutationCache in pkg/controller/resourceclaim/controller.go (around line 356) is configured with IncludeAdds: true and a 1h TTL, but the claim informer handlers never call OnAddOrUpdate/OnDelete on it. Per the report, after a generated claim is deleted and the informer has observed the deletion, claimExists and findPodResourceClaim still return the cached entry, so the controller does not recreate the claim until the entry expires. The report ties claimExists to #140227 and says it is in v1.37. The reporter reproduced this at commit 4dfdf5067a2 using a unit test (mutate, delete via API, wait for informer, claimExists returns true). No exact error string is given; the symptom is delayed recreation, not a logged failure. Cause status: The reporter's explanation (missing informer event wiring into the mutation cache) is an unconfirmed hypothesis plus a reporter-run test. No maintainer has confirmed it; the only visible comment is the automated triage bot saying the issue awaits triage (https://github.com/kubernetes/kubernetes/issues/142870#issuecomment-6080483685). Possible direction (not a verified fix): The reporter says wiring the informer events the way #140063 did for the ResourceSlice controller fixes the problem. No PR or release is linked in the supplied sources, so fix status is unknown/unshipped. No workaround is stated. Diagnostic hint (conditional on the report): if on a build including claimExists a deleted generated claim is not recreated while the informer already shows it gone, a stale mutation cache entry is a candidate explanation; waiting for TTL expiry is implied by the report but not given as a recommended workaround. Limitations: single reporter, unit-test evidence only, partial comment coverage, no maintainer confirmation, no fix or release evidence, and no KFA reproduction.