Add rule no-sibling-hooks (fixes #82)

[jfmengels]

Add rule no-sibling-hooks (fixes #82)

Let me know if you want me to change something or if I've missed something :)

I deactivatd two rules here about the function complexity. I could not make the function less complex. I could move the last part of the function to another function, but I'm not sure it would make it less complex, as you'd pass a few arguments around. Not opposed to doing that though. Let me know what you wish for me to do.

[lo1tuma]

-but it can confusing
+but it can be confusing

[jfmengels]

Additionally to that should we also support things like xdescribe, xcontext, describe.skip, describe.only?

Added support for those

Should we also support mocha’s TDD interface which uses suite instead of describe and maybe also the describe alias context?

Still TBD. I'd vote for no unless someone really wants it or it's properly handled with the other rules. If this is the only rule that will support it, I don't see the point.
EDIT: I get the feeling that it actually is. So up to you @lo1tuma to decide :)

[jfmengels]

Actually, if we want to do that, we'd probably better update no-hooks and similar (existing or upcoming rules), but I'm a bit worried that setup for instance is a too common function name, and disallowing that one in no-hooks would be an unwanted effect. It would be fine if we can know for certain whether the linted file is a test file or not, but I wouldn't know how to do that.

Or actually, we could probably do it by not reporting the variables that were declared in the file (using context.getScope(). Unless setting mocha: true in the ESLint env settings makes them available in the scope. Will try to play with this to know if we can have a nice solution.

[lo1tuma]

I would be ok with not supporting the TDD interface now. We can add support for it if somebody requests that.

I think the right approach would be check where setup and friends were declared (might be also imported with custom name) or if it is a global. But as I said, we can skip this for the moment and wait until somebody reports such a use-case.

[jfmengels]

I'm fine with that. Then this PR is up for review

[lo1tuma]

LGTM

[jfmengels]