Mechanism to force clients to re-hit build-info endpoint on demand

[AndyGrant]

Needed when out-of-sync updating instances
Needed when making minor instance changes, without trying to kick the version

The act of restarting the webserver should cause currently-active engines to restart the pipeline. Unclear how to do this without modifications to client.py. May need to create a new exception in the client, proc it in a workload request endpoint, and use it to reupdate info as an except block in the main loop of the worker.