feat!: Drop node.js 6, upgrade dependencies

[coreyfarrell]

No description provided.

[coveralls]

Coverage decreased (-0.01%) to 96.842% when pulling 6bc104f on coreyfarrell:node8 into 421c5bd on istanbuljs:master.

[lukeapage]

(edit) I don't think its a bug in this PR. in fact I have no idea why I hit the problem, seems for me I need to specify the path directly in spawn sync rather than just rely on it knowing nyc. Maybe I don't have nyc installed globally? Sorry for the confusion.

---

I don't know if its related to this change or istanbul, but I made instructions to test the master of the instabuljs mono-repo with it dog-fooding its own source, for trying out report changes:

https://gist.github.com/lukeapage/c7844b3bd7fab7eed260bfef387848b9

After fixing the none report bug though, it still doesn't work.

It looks like the merge command on this pr as run here:
https://github.com/istanbuljs/istanbuljs/blob/master/monorepo-merge-reports.js#L19

doesn't merge, but generates the reports as it might do with normal report generation.

[bcoe]

this is looking good to me, as soon as those Node 8 tests are passing 👍

spread, how fancy.

[JaKXz]

👍 from me; like Ben said, when the tests are passing it's LGTM!

[coreyfarrell]

The issue is not node.js 8, it's Windows (yes, how shocking). I suspect the problem is due to istanbuljs/istanbuljs#422, not sure if this is exposing a failure in test-exclude or if the nyc tests are just wrong for Windows. It's unlikely I'll be able to work on this before next week.

[coreyfarrell]

Finally got around to dealing with the test. Basic issue is that testExclude.shouldInstrument no longer accepts / based paths on Windows, the root path needs to be drive based (thus C:// or similar). This is a non-issue in practice, the problem is that when the test ran nyc.exclude.shouldInstrument('/cwd/foo.js', 'foo.js'), this is something that would never happen in real life.