ClientError: ensure errors don't bork reconnection attempts #762
Description
An issue where the Cylc UI Server may require restart to reconnect to a workflow.
We recently encountered a situation where a workflow host died. Workflows which were taken out by this were restarted on other servers, however, they did not update in people's Cylc GUIs.
The Cylc UI Server had received an error along the lines of "workflow was running on <host-1> is now running on <host-2>". This error appears to have been terminal for the UIS connection to this workflow, it does not look like the UIS attempted to reconnect at a later date.
User requests for workflow scans did not resolve the issue. The UIS had to be restarted.
This may be a related to / a generalisation of #754