Add support for service health checks for the Kubernetes executor

What does this MR do?

It adds code to make the kubernetes executor wait for services to start on their specified HEALTHCHECK_TCP_PORT (if one is specified) before the executor starts any build jobs.

Why was this MR needed?

Typically build or test jobs depend on the availability of the services, but without this change there's a risk that they might not have finished starting up.

What's the best way to test this MR?

Configure a project with services, specifying HEALTHCHECK_TCP_PORT. We log lines to the job log saying "waiting for TCP connection to [service]". Ensure that all configured services are waited for on the appropriate ports.

What are the relevant issue numbers?

close #27215 (closed)