The remote_contract job was effectively dead code — it only ran on
workflow_dispatch (excluded push and pull_request events) and was
configured with continue-on-error, so it never blocked releases.
Removing it simplifies the pipeline and eliminates the always()
workaround in the release job's if-condition.