Reorganize build/test in CI

[ss-es]

No linked issue.

This PR:

This is a PR to reorganize CI. Changes:

• The self-hosted build, build-release and lint jobs are removed. It is unnecessary to duplicate these on GitHub runners and on the self-hosted runners. The self-hosted test and build-arm jobs are retained.
• The build job now depends on test passing. This means that tests run first, and that we never build if tests fail. In particular, we still do not build or push dockers if the tests fail.
• The build-release job runs in parallel to tests, since the artifacts from this job were not used to build dockers.
• We no longer skip test_crypto in the regular tests, so it no longer runs as a separate job.
• We no longer cancel the async-std build or test if the tokio variant fails, and vice-versa.
• The self-hosted test no longer runs with --no-fail-fast. This means test failures are reported (in the form of CI failing) immediately, which should be helpful for development. This doesn't affect the regular GitHub runners, which will still continue to run all tests even after encountering a failing test, and may be helpful for debugging. (The choice of which one to run with --no-fail-fast was completely arbitrary.)

Unrelated to the primary changes, this PR also adds DEBUG logging for libp2p-networking, which is helpful when the tests in libp2p-networking/tests/counter.rs fail -- currently, these tests don't produce any useful logs when they fail.

Notes: if we expect build-release artifacts to be consumed in the future, it should depend on passing tests.

This PR does not:

Key places to review:

Overall test workflow, e.g. here: https://github.com/EspressoSystems/HotShot/actions/runs/8348230729?pr=2734

[jparr721]

🥳