Fixed retry and duplicated replacements

[Seetaramayya]

For good amount of time our scala-steward jobs are failing. Following changes fixed in our case, so creating the PR, might be useful port the changes to the main repo.

I hope this PR address the issue #2934

New Retry logic

• If Retry-After header present, then use the header value to retry (this was there before)
• otherwise retry request after time specified by X-Ratelimit-Reset header. Header value is UTC epoch seconds. Here I took some assumptions
  • if the difference between header value and current UTC epoch seconds is more than 60 seconds then I retry after 60 seconds. That means max wait time would be 60 seconds.
  • if the difference is negative then I retry after 1 second. That means min wait time is 1 second
• if neither of the headers present, wait 1.second

Here is the github suggested best-practices

Another issue

If there are duplicated replacements then the job was failing with StringIndexOutOfBoundsException. Added unit test to verify the behaviour.

[fthomas]

Another issue

This deserves its own PR. I would also prefer to have a test for that in RewriteTest instead of in EditAlgTest.

[Seetaramayya]

Another PR crated to address the other issue #3016

[Seetaramayya]

Are you going to merge this PR?

[odisseus]

In order to make this PR mergeable, you must resolve the conflicts first.