Potentially redundant/upliftable lints

[clarfonthey]

A few lints I noticed which may be redundant to newer lints in rustc upstream:

• absurd_extreme_comparisons should be uplifted to the existing unused_comparisons
• bad_bit_mask should probably be uplifted to the existing unused_comparisons lint, or an unused_bitmask lint
• bool_comparison should probably be uplifted to unused_comparisons
• cmp_nan should probably be again uplifted to unused_comparisons
• eq_op should probably be again uplifted to unused_comparisons
• unused_collect should be replaced by the existing must_use on collect
• useless_attribute should be replaced by unused_attribute
• out_of_bounds_indexing should be a const eval error
• iterator_step_by_zero should be removed; this now panics
• enum_clike_unportable_variant should be part of a compiler portability lint

(will make larger list when I get around to it)

[Manishearth]

FWIW collect already has must_use, this must have happened in the recentish flurry of PRs adding must_use everywhere. So we can remove this lint if nobody objects.

As for absurd_extreme_comparisons, the clippy lint is better, we should probably uplift the goodies we have.

[clarfonthey]

So as I'm going through a bunch of lints… there are a lot of lints which could just be added to the existing, upstream unused_comparisons lint. A lot of these are super easy to check, with a huge net gain for checking. Plus, I can't see why someone would want to disable them individually, considering how they're very straightforward.

Perhaps the bad_bit_mask lint would be a bit too much for upstream rustc, but a lot of the others would be very good to add upstream.

[nk9]

Seems like this issue either needs to be updated, split up, or possibly just closed. Hopefully someone with more knowledge of the project's history and existing lints can look at this with a fresh pair of eyes.

@rustbot label S-triage C-an-interesting-project