Test failures in Ubuntu with Python 3.10

[davidorme]

Having said that, and totally underelated, the tests with Ubuntu and Python 3.10 are a bit flaky, right? They fail for no apparent reason and pass if re-run. Any idea of why that might be?

Originally posted by @dalonsoa in #384 (comment)

They sure are flaky - initially this looked like some kind of transient issue but over on #384 it is failing consistently. It isn't clear what is going wrong - partly because the error reporting within vr_run is too monolithic and partly because we can't access the log from the Github Action runner at the moment.