feat: use thread pool for trigger job queue to prevent overloading.

[chrisLeeTW]

This PR try to fix this issue #162 which use thread pool for trigger job queue to prevent overloading.

And this strategy is also used by Github pull requests builder plugin (https://github.com/jenkinsci/ghprb-plugin) & Stash pull request builder plugin (https://github.com/jenkinsci/stash-pullrequest-builder-plugin)

I already test this patch in our company production jenkins server for 2 months without problem.

[CodeMonk]

@chrisLeeTW So, first off, I love all the cleanup - things look, in general, much cleaner and more readable.

But, I want to make sure that this is necessary. Do you believe that #162 is not the same thing as the timer stuck bug, where things stop polling completely? I had assumed, possibly incorrectly, that #162 was the same issue as #180

Let me know ASAP. I want to push a release soon (today), and I'd like to get this cleanup and/or fix.

-Dave

[chrisLeeTW]

@CodeMonk I have some comment on PR (#183).

And I have to let you know that we are back to use the bitbucket cloud v1 api endpoint in current master branch, which I already fix in this pull request (#159), so all things is broken for bitbucket cloud part.

So before this regression is fixed, I think you may not release a new version. 😢

[chrisLeeTW]

just for reference, my idea has came from this pull request (nemccarthy/stash-pullrequest-builder-plugin#147)

[CodeMonk]

@chrisLeeTW Which regression - where did v1 come back?

[CodeMonk]

@chrisLeeTW Please review: jenkinsci#20