Fix Long overflow in Retry when backoff factor > 1.0

[dlmarion]

Closes #3414

[EdColeman]

This may prevent the overflow, but I was reviewing this to see if the wait / backoff factor is what is expected when this was merged. Where

currentWait = Math.min(maxWait, initialWait + waitIncrement);

should it be something like

currentWait = Math.min(maxWait, currentWait + waitIncrement);

If you have an initial wait of 10 seconds and then you want to retry every second with a back off I would expect wait times like 10, 11, 13, 17,... But it appears that the code would be like 10, 21, 33 (or something), but basically backing off by the initialWait and the back-off increment.

[dlmarion]

So, two bugs in our Retry object then? Maybe if we are going to touch this again, instead of fixing it we rip it out and replace it with one of:

https://curator.apache.org/apidocs/org/apache/curator/retry/ExponentialBackoffRetry.html
http://rholder.github.io/guava-retrying/
https://resilience4j.readme.io/v1.7.0/docs/retry