Give the CI fuzzer more time

[nathaniel-brough]

Currently the CI fuzzer isn't on the critical path to completion. So we can extend the amount of fuzzing time without impacting the critical path.

[nathaniel-brough]

For context the "ci / test (pull_request)" seems to be the critical path.

[nathaniel-brough]

Ah, it looks like I increased the time too much. I'll drop 40s off this tomorrow.

[Byron]

Thanks for the proposition. However, seeing the fuzzing stage to now dominate the runtime, I'd like to not pursue even if the runtime was reduced by a minute.

Instead, I'd like to see the current difference between CI-test and CI-fuzz as safety-buffer. The reason for not wanting to maximise fuzz time is the unpredictability of CI runtimes in general, while experience shows that the fuzzer finds obvious errors even within a very short time, while taking much longer than the time we have for follow-ups.

Thanks for your understanding.