bump istio, go-control-plane and k8s 1.32 deps

[stevenctl]

Description

• Bump go-control-plane to latest, including the move to multi-module
• Drop OpenCensus support since envoy dropped it
• Bump istio/istio dep since it's tied to envoy go-control-plane
• Work around the fact that newer istio brings async krt processing.

API changes

• Drop OpenCensus support since envoy dropped it. We will now return "unsupported" for this config in our own API.

Code changes

• Bump go modules istio and go-control-plane
  • ef139ef8ea6b06cf0424f37883f50ccf95743c51 of github.com/envoyproxy/go-control-plane because it has upstream_proxy_protocol: Introduce custom TLV support envoyproxy/envoy#37591
• Change filter_types gen to include the various envoy submodules. Some imports are duped but that should be ok.
• Remove OpenCensus references that don't exist in newer go-control-plane
• Add a check that prevents us from sending xds snapshot until we have processed per-client clusters. This will never send xDS if there are 0 upstreams, which likely never happens even in tests.
• Bumping fsnotify would introduce a race in the sds filewatching test: Race when calling Add from a goroutine fsnotify/fsnotify#672
• Made ggv2setup_test less flaky by ensuring endpoint pods are ready before we request xDS.
• KRT changes:
  • Our Settings collection must be a *Settings collection now because only the pointer properly implements controllers.Object which lets krt generate keys for it.
  • krt Get no longer needs krt.Key and instead takes a raw string; this needed small updates in several places
  • Rather than calling Syncer().HasSynced(), Collection now embeds the Syncer interface.

Docs changes

Context

Interesting decisions

Testing steps

go test ./projects/gateway2/setup/ggv2setup_test.g specifically will fail
the destination rule/failover cases without the perClientClusters == 0 check
and passes when adding it back.

Notes for reviewers

Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works

[stevenctl]

hmm some failure with kuztomize conflicting, will look more tomorrow.

[nfuden]

nit: extra whitespace changes bundled in to prs make them harder to read

[stevenctl]

We should have standard auto-formatting that is checked in CI to avoid this stuff..

[solo-changelog-bot]

Issues linked to changelog:
kgateway-dev#10487

[Nadine2016]

could we add a description here?

[stevenctl]

Hmm. We don't have any description for the pod security context stuff and it likely is just a mirror of kuebrnetes fields.

[Sodman]

Approving to unblock because changelog is the only thing that has changed since the prior reviews.