fix: ensure connectivity with Kubernetes API on start-up

[czeslavo]

What this PR does / why we need it:

It introduces a Kubernetes API status check on start-up to prevent issues like #4603 and #4207. With the connectivity ensured, we may rely on controller-runtime default behavior (cache sync timeout will be triggered in case of missing CRDs) and drop our manual checks for Kong's CRD existence.

Adds envtests:

• TestManagerDoesntStartUntilKubernetesAPIReachable to ensure the manager will not start its runnables (including the config sync loop) until the Kubernetes API is reachable.
• TestNoKongCRDsInstalledIsFatal to ensure that in case of missing Kong CRDs installation, the controller will eventually crash.
• TestGatewayAPIControllersMayBeDynamicallyStarted to ensure Gateway API CRDs can be installed during the runtime and their controllers will be started dynamically in such a case.

Also, migrates Knative Ingress controller to use DynamicCRCController just to not maintain 3 ways of initializing the controllers.

Which issue this PR fixes:

Should fix #4603 and #4207.

PR Readiness Checklist:

Complete these before marking the PR as ready to review:

• the CHANGELOG.md release notes have been updated to reflect any significant (and particularly user-facing) changes introduced by this PR

[codecov]

Codecov Report

Patch coverage: 85.4% and no project coverage change.

Comparison is base (0e04e18) 68.0% compared to head (13c68b0) 68.1%.
Report is 2 commits behind head on main.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #4641   +/-   ##
=====================================
  Coverage   68.0%   68.1%           
=====================================
  Files        163     163           
  Lines      19091   19067   -24     
=====================================
- Hits       12992   12987    -5     
+ Misses      5331    5304   -27     
- Partials     768     776    +8     

Files Changed	Coverage Δ
internal/cmd/rootcmd/run.go	19.0% <0.0%> (-2.1%)	⬇️
internal/manager/conditions.go	89.6% <ø> (-2.1%)	⬇️
internal/controllers/crds/dynamic_controller.go	73.8% <33.3%> (+56.3%)	⬆️
internal/manager/run.go	55.1% <69.4%> (+1.8%)	⬆️
internal/manager/controllerdef.go	98.7% <100.0%> (-0.2%)	⬇️
internal/manager/setup.go	75.5% <100.0%> (ø)

... and 3 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rainest]

I'd had some concern that OpenShift would do something non-standard for its readiness endpoint, but https:/openshift/installer/blob/master/docs/dev/kube-apiserver-health-check.md confirms it's still /readyz, and building the URL from GetConfig() will deal with their using something other than the kubernetes Service.

We may want to avoid testing the missing CRD cache timeout behavior, since I don't think it's a guaranteed controller-runtime behavior. They do test it upstream, though with a fake cache that simulates no-CRD behavior rather than actually not installing something.