What type of PR is this?

/kind feature
/kind cleanup

What this PR does / why we need it:

Creates a dedicated e2e/waiter package that provides a single, generic Waiter[T] struct driven by user-supplied fetch functions and predicates. This replaces the scattered wait/poll
logic that was previously spread across k8s.PollForCondition, manager methods, and inline closures with inconsistent signatures.

Key changes:

• e2e/waiter/waiter.go — Generic Waiter[T] struct with builder-pattern configuration (New[T]().WithTimeout().WithInterval().WithLogger().WithRetryOnError()), WaitFor (returns value) and
  WaitUntil (discards value) methods, typed FetchFunc[T]/Predicate[T] types, built-in predicates (IsNotZero, IsZero, AlwaysTrue), helper converters (ToFetchFunc1, ToFetchFunc2,
  FetchByName), and resource lifecycle helpers (WaitForResource, WaitForResourceDeletion)
• e2e/k8s/pods/predicates.go — Reusable pod predicates: AllReady, AllPending, MatchPhases, ReadyCount, CountEquals, ExcludesPod, HasUnschedulableEvents
• Migrated all ~30 call sites from k8s.PollForCondition and wait.PollUntilContextTimeout to the new waiter package with properly typed fetch functions and predicates
• Deleted e2e/k8s/polling.go — the old PollForCondition is fully replaced
• Removed Option functional-options pattern in favor of builder-style method chaining
• Removed PodWaiter()/FetchPods()/WaitForCondition() from TestContext — all pod wait logic now delegates to PodManager methods

Which issue(s) this PR fixes:

Fixes #518

Special notes for your reviewer:

• Every waiter uses the natural K8s type of what's being fetched (e.g., *v1.PodList, *PodCliqueSet, *v1.Node, *v1.Secret) rather than bool. This eliminates the unsafe closure-capture
  pattern where callers declared var result T outside PollForCondition and assigned it inside.
• WaitForResource and WaitForResourceDeletion are free functions (not methods) because they require T comparable for IsNotZero/IsZero, which can't be added as a constraint on Waiter[T
  any] methods.
• NotFound handling is now explicit in each fetch function rather than blanket WithRetryOnError() — this preserves the original behavior where NotFound keeps polling but other errors
  fail immediately.
• The waiter package now imports k8s.io/apimachinery for GetFunc[T] and errors.IsNotFound.

Does this PR introduce a API change?

NONE

Additional documentation e.g., enhancement proposals, usage docs, etc.:

See e2e/waiter/waiter.go for the full API. Usage example:

  // Wait for a resource to exist
  w := waiter.New[*v1.Node]().
             WithTimeout(timeout).
             WithInterval(interval).
             WithLogger(logger)                                                                                               
  return waiter.WaitForResource(ctx, w, nodeName, clientset.CoreV1().Nodes().Get)                                                                                                          
                                                                                                                                                                                           
  // Wait with custom predicate                                                                                                                                                            
  w := waiter.New[*v1.PodList]().
             WithTimeout(timeout).
             WithInterval(interval)                                                                                                               
  pods, err := w.WaitFor(ctx, pm.FetchFunc(ctx, ns, selector), pods.AllReady(expectedCount))       
  
    // Wait with custom predicate                                                                                                                                                            
  w := waiter.New[*v1.PodList]().
             WithTimeout(timeout).
             WithInterval(interval)                                                                                                               
  err := w.WaitUntil(ctx, pm.FetchFunc(ctx, ns, selector), pods.AllReady(expectedCount))                                                                                                  
                                                                                                                                                                                           
  // Wait for deletion                                                                                                                                                                     
  w := waiter.New[*PodCliqueSet]().
             WithTimeout(timeout).
             WithInterval(interval)                                                                                                             
  return waiter.WaitForResourceDeletion(ctx, w, name, client.PodCliqueSets(ns).Get)