fix(docker): exec the compose command instead of always starting the server

[Eljees]

Fixes #544

docker/entrypoint.sh ends with a hardcoded server start and never looks at its arguments, so the
command: from docker-compose is ignored by every container built on this image. In the default stack
that means faraday-default-worker (compose declares command: [ "faraday-worker" ]) starts a second
server instead of a Celery consumer:

Starting Faraday server ...
WARNING - Faraday Server is already running. PID: 21

and every report uploaded via CLI/Web UI/API queues in Redis forever (redis-cli LLEN celery keeps
growing, celery inspect ping → "No nodes replied").

The entrypoint now execs the passed command when there is one, and keeps the current behaviour when
there is none:

if [ "$#" -gt 0 ]; then
    echo "$(date) Executing command: $*"
    exec "$@"
fi

echo "$(date) Starting Faraday server ..."
exec faraday-server --bind 0.0.0.0

exec also makes PID 1 the actual service in both cases, so signals from docker stop reach the
server/worker instead of the shell.

The init part of the script (server.ini creation, migrations) still runs for the worker as before —
the worker starts with depends_on: service_healthy, so it takes the "Using existing server.ini" branch;
this PR does not change that behaviour, it only stops the worker from turning into a second server.

Verified

bash -n docker/entrypoint.sh clean, plus a stub run (fake faraday-server/faraday-manage/nc on PATH,
existing server.ini):

with a command:    Executing command: echo I-AM-THE-WORKER   ->  I-AM-THE-WORKER
without a command: Starting Faraday server ...               ->  faraday-server --bind 0.0.0.0

AI-assisted (LLM used for drafting); the change and the runs above are mine.