For anyone else seeing this issue, support confirmed it was because all proxies were busy with backup jobs. I was baffled that if a proxy was required for restore, that there wasn't at least 1 dedicated to restore tasks. Support took my feedback as a feature request for a future version.
We had to cancel all queued and at least 1 running job to free up a proxy in order to perform the restore. This can be done using mccli via cli on the VDP appliance. Run mccli activity show to display running and queued jobs, then mccli activity cancel --id=<id> to cancel each job, then we could perform the restore and lastly had to re-issue backup jobs. Certainly not something I can ask my system admins to be doing every time a restore is needed.
I agree about VDP. It's got so much potential, but I'm nearly at my wits end with it. VDP 5.5.5 is better than all previous versions, but still requires WAY too much babysitting.