Fix poll_retry and add max_num_poll_retry

[annesylvie]

Hi!

I was looking at the logic that handles retries when a service/node is dead, and I noticed an inconsistency between what the retry_poll does in the documentation and how it gets used - it looks like poll_retry acts as a maximum number of retries to do rather than a duration to wait between two retries.

I've added some minor changes to make the code reflect the documentation and added a new max_num_poll_retry config as I know it would be useful for us at least.

I am a bit unsure if this is fully a good idea as the user could set a long sleep time which would block probing other nodes in the meantime, potentially for a long time.

Let me know what you think, I hope this is all correct and not a misunderstanding on my part! Thanks.