TF request retries

[The-Mule]

Original EWA babysits executed test jobs and attempts to handle errors. E.g. when a job aborts it reties it (it does so using beaker-jobwatch tool). If possible we can try the same approach somehow.

[kkaarreell]

I think there are at least two associated use cases.

1. babysit running tests jobs and reschedule them if necessary - this should be possible in the current code by adding an extra code to the "execution" loop.
2. ability for a user to re-run already finished job - for that I do not have a good answer, except that we should be able to point user to the respective TF job so it can be rerun. Also, if newa is executed locally it should be possible to restart from a certain part of the pipeline based on the present yaml files.

@The-Mule I read this issue that this is rather about 1. if not, feel free to open another issue describing the missing pieces.