(Docker 2/5) Label suspended docker-autoscaler environments for TTL cleanup

What does this MR do?

Labels the containers, networks and volumes of a suspendable docker-autoscaler job with the runner ID, acquisition key and expiry (expires-at, creation plus runtime_environment_ttl). Suspend and resume are refused when the runner could never clean the environment up.

Why was this MR needed?

The labels let the runner find an environment later and check its expiry on resume. Nothing is resumed differently or deleted yet.

What's the best way to test this MR?

go test -race ./executors/docker/ ./executors/docker/internal/labels/

What are the relevant issue numbers?

Related to #39468.