PLA 509 dump docker compose logs on failure (dagster-io#23167)

## Summary & Motivation

when docker-compose up throws an exception, just dump docker logs to
stdout. sometimes there isn't anything more revealing here than what we
get in the initial exception, but sometimes it points to an issue with
the underlying service that wasn't in the original exception

## How I Tested These Changes

forced docker-compose to fail with a bad health check definition, and
ensured that the dump_logs function was invoked and did print to stdout

---------

Co-authored-by: Neil Fulwiler <[email protected]>

python_modules/dagster-test/dagster_test/fixtures/docker_compose.py

@@ -21,7 +21,14 @@ def docker_compose_cm(
     if not network_name:
         network_name = network_name_from_yml(docker_compose_yml)
     try:
-        docker_compose_up(docker_compose_yml, docker_context, service, env_file, no_build=no_build)
+        try:
+            docker_compose_up(
+                docker_compose_yml, docker_context, service, env_file, no_build=no_build
+            )
+        except:
+            dump_docker_compose_logs(docker_context, docker_compose_yml)
+            raise
+
         if BUILDKITE:
             # When running in a container on Buildkite, we need to first connect our container
             # and our network and then yield a dict of container name to the container's
@@ -36,6 +43,21 @@ def docker_compose_cm(
         docker_compose_down(docker_compose_yml, docker_context, service, env_file)
 
 
+def dump_docker_compose_logs(context, docker_compose_yml):
+    if context:
+        compose_command = ["docker", "--context", context, "compose"]
+    else:
+        compose_command = ["docker-compose"]
+
+    compose_command += [
+        "--file",
+        str(docker_compose_yml),
+        "logs",
+    ]
+
+    subprocess.run(compose_command, check=False)
+
+
 @pytest.fixture(scope="module", name="docker_compose_cm")
 def docker_compose_cm_fixture(test_directory):
     @contextmanager