require macro and generic testcases for new lints and lint updates

[gnzlbg]

Given how many lints are currently broken for code inside macros, and how many lints have been so over the years, I think it should be a requirement for new lints to at least contain one test that tries to break the lint when used inside a macro. I believe most issues could have been caught by just remembering that macros exists when testing a new lint.

[gnzlbg]

I also believe that the same applies to generics.

Also, I am not suggesting that new lints or updated lints should "work" on macros or generics. The only thing that I am suggesting is that if they cannot work in these contexts, they should try to detect these and bail out silently instead of displaying false positives.