Improved support for numeric brace ranges

[koopero]

Description

This PR improves support for numeric brace ranges in micromatch globbing ( ie {1..9} ). The top-level methods micromatch, isMatch, makeRe and capture are modified to use the newly introduced braces module. New test cases have been introduced to cover these additions.

This PR also introduces some disabled test cases which cover errors in the upstream braces module, specifically bugs with conversion of zero-padded brace range to regex ( ie {000...200} ).

Since this functionality is already covered in docs, no changes have been made to README. Version has been bumped to 4.0.3, reflecting a patching of existing functionality.

[jonschlinkert]

@koopero sorry for the late reply. I've been mulling this over. I might prefer a different approach to solving this in micromatch, but first I think it will help to get the problem fixed in underlying libs, so I'll review your other PRs, then we can come back to this one. Thanks for doing this!!!

[koopero]

@jonschlinkert No worries. I haven't had time for this collection of issues either. Thank you for your patience and hard work.

At the very least, I don't see merging this until we take care of the upstream issues in braces. As well, there is another bug in makeRe and capture, specifically handling of single char ? glob patterns.

I think you should consider how the capture and makeRe functions fit into micromatch vs other matching modules in your repertoire. To my mind, these features should work perfectly or not be included.

I am happy to help with improvements, but it would be helpful to have more context as to the scope, intentions and connections of modules with the micromatch network.

Cheers!