Backoff unittests are fragile

[eccles]

Describe the bug
The unittests test the backoff logic by setting a low timeout value. This leads to test fragility - sometime one has to rerun the github action when submitting code or releasing.

To Reproduce
Monitor actions - sometimes the action fails - rerunning usually fixes it.
The action fails because test coverage is less than 100%

Expected behavior
Never fails.

Additional context
One should 'mock' the backoff logic - this is a decorator so some difficulty with this

[eccles]

The unittests set a low max_time of 1 second and than manipulates the responses to return data such that the backoff logic exits at the right moment. The difficulty is that one has no way of predicting the individual sleeps that make up the backoff creiteria.

For example a uniittest will testan unconfirmed response by scheduling (say) 6 responses non of which will cause a successful confirmation. Most of the time this works but it is possible that the 6 responses will not collectively amount to 1 second...

Conversely tests for confirmation involves responding with a response that will terminate the backoff successfully - occasionally the first backoff sleep may exceed 1 second - solution is to set max_time to a large value perhaps...