With#5799 we can now restart workspaces, however, the restart process is dependent on the client sticking around and issuing the start command after stop has completed.

We should implement restart functionality in the API so that we're not dependent on the client.

We could also consider creating retry-functionality so that if either stop or start fails, it can be retried a few times.

It'd be nice if this were implemented in the backend as a job type long-term... maybe we open an issue for it.

Originally posted by@kylecarbs in#5799 (review)