compute: remove linkerd workarounds from networking

[benesch]

This essentially reverts #12832, which I suspect is no longer necessary now that we no longer use linkerd in Materialize Cloud. The easiest way to find out is to try it while we're still in alpha. We can always revert this PR if we rediscover the networking issue.

For good measure, this commit also removes the handshake magic, which is not necessary now that timely communication is not multiplexed on the same port as HTTP and SQL traffic.

Motivation

• This PR removes a now-unnecessary workaround.

Checklist

• [x] This PR has adequate test coverage / QA involvement has been duly considered.

• [ ] This PR evolves an existing $T ⇔ Proto$T mapping (possibly in a backwards-incompatible way) and therefore is tagged with a T-protobuf label.

• [ ] This PR includes the following user-facing behavior changes:

  • n/a

[umanwizard]

Oh, actually, would you mind checking Honeycomb to see if we're seeing failed computed handshake with worker in the logs, before you land this? If we have never seen it (or not since linkerd was ripped out), that would be solid evidence that the issue this was designed to prevent isn't reproing under the new cilium regime.

[benesch]

Oh, actually, would you mind checking Honeycomb to see if we're seeing failed computed handshake with worker in the logs, before you land this? If we have never seen it (or not since linkerd was ripped out), that would be solid evidence that the issue this was designed to prevent isn't reproing under the new cilium regime.

Done. No trace of any such messages in Honeycomb!