You could start another "empty" instance of a runtime (easy when using containers/Docker)
You could save/deploy the flows in de-activated state before exiting the runtime, so you can manually activate them as needed after restarting the runtime.
For point 1 click on the dropdown arrow on the Deploy button, there you will find Restart Flows which may be what you are looking for.
For point 2, as has been suggested by others, stop node red then start it again using node-red --safe
which will start the editor but not the flows. If you make changes then Deploy will start the flows, or you can use Restart Flows as for point 1.