Correct signedness handling of DispatchTimeInterval

[dgrove-oss]

Replicate e32016ffd765020c696bc5d1e42606074305f2db
from the Darwin overlay to the wrapping overlay.

[dgrove-oss]

Do we need a swift 3 branch version of this patch as well?

[mwwa]

This looks good, there's no need to port to the swift repo as the changes are already there. Is that what you meant? (Or do you mean a swift 3 branch of libdispatch?)

[mwwa]

@swift-ci please test

[dgrove-oss]

I meant the swift3 branch of libdispatch. I think this is an internal implementation change that doesn't impact the exposed API, so we don't need it in swift3-branch, but I wanted to confirm that with you.

[mwwa]

Gotcha, yeah that's what I thought too.

[dgrove-oss]

The PR test failed with an assertion in libkqueue when running a Foundation test case. I've built and run locally half a dozen times and can't reproduce. It's possible I'm missing something, but it might have been a glitch on the test machine. I did notice at the end of the log that Jenkins complained:

[xUnit] [ERROR] - Clock on this slave is out of sync with the master, and therefore 
I can't figure out what test results are new and what are old.

[mwwa]

Ok, I'll go ahead and merge.