feat: multi cluster resource management and cache efficiency

[zachsmith1]

This pull request introduces several improvements to multicluster admission handling and configuration, focusing on cluster-aware resource management, cache efficiency, and code maintainability. The most significant changes include switching cluster identification from annotations to labels, adding efficient client/informer pools for per-cluster operations, introducing a CEL runtime cache for webhook admission, and ensuring proper decoration of REST options for multicluster routing.

Cluster-aware resource handling and configuration:

• Cluster identification is now managed via labels instead of annotations in both mutating and validating admission plugins, ensuring consistency and better alignment with Kubernetes conventions.
• Per-cluster client and informer pools are introduced and wired through the configuration and namespace manager, enabling efficient reuse and sharing of cluster resources for admission and lifecycle plugins.

Multicluster REST options decoration:

• A new decorateRESTOptionsGetter helper ensures that REST options getters are properly decorated for multicluster routing across controlplane, apiextensions, and aggregator servers, with logging for decoration status.

Webhook admission improvements:

• A new CEL runtime cache (CelRuntime) is implemented for webhook admission, providing efficient, per-cluster and per-resource CEL environment compilation and metrics for cache hits/builds. Associated unit tests verify caching behavior.

Namespace lifecycle and plugin enhancements:

• The namespace lifecycle plugin now avoids deadlocks on initial cache sync by setting a ready function that always returns true, improving reliability during cluster initialization.