Unfixed coverage version to use latest one - #132

This looks good, but my one question is how to test the problems with previous 4.x versions have gone away... Should we just merge and see if any affiliated package starts complaining? This may be the only feasible way, but then perhaps it is a good idea to send a note to the mailing list...

Alternatively, but maybe too much work, if one can override the package (or reinstall it) in .travis.yml, we could try it in astropy/astropy#5308

My travis-fu is not that strong. Perhaps @astrofrog or @bsipocz can suggest a way...

I recall that there were some previous problems, that's why we froze the version numbers here. So I suggest to double check it with a few mock branches for at least astropy core (and maybe a few affiliated packages like gammapy that saw the issue) before merging this. Generally we try not to experiment much in the master ci-helpers is avoidable by doing testing first.
I'll try to have a look at this later tonight.

Thanks, @bsipocz !

@pllim - Could you please rebase this? (As a safeguard PRs in ci-helpers always needs to be against the latest master, so whenever the master changes everything needs to be rebased and retested. before github allows to push the merge button.)

@bsipocz , rebased and passed.

It looks like that coverage runs with the new version, both in astropy and gammapy, so I'm happy to merge this asap.
(Although I get some failures in the test builds, I think those are unrelated and should go away once the build is run in its repository and not in my fork: https://travis-ci.org/bsipocz/astropy/builds/166497095)

@pllim - Could you please rebase it one last time?

@bsipocz , rebased and passed.

Hurray, thanks @pllim.