Add a lint to warn on `T: Drop` bounds

[notriddle]

What it does: Checks for generics with std::ops::Drop as bounds.

Why is this bad? Drop bounds do not really accomplish anything.
A type may have compiler-generated drop glue without implementing the
Drop trait itself. The Drop trait also only has one method,
Drop::drop, and that function is by fiat not callable in user code.
So there is really no use case for using Drop in trait bounds.

Known problems: None.

Example:

fn foo<T: Drop>() {}

Fixes #3773

[oli-obk]

Idk how hard it is to create a structured suggestion for this lint, but I'd be delighted if you tried to create one. Maybe modify a copy of the AST in place and pretty print that?

[notriddle]

I don't think suggestions would help very much. While "just remove the bound" would "fix" the code in the sense that it would stop the lint from firing and it wouldn't break anything that worked, the proper way to fix it is to switch to std::mem::needs_drop instead, and I am not going to try to generate that kind of massive refactoring.

[notriddle]

I also just realized that I hadn't actually tested this with where clauses. I should probably fix that...

[oli-obk]

r=me with docs nit and travis passing

[notriddle]

Okay, @oli-obk. Fixed the formatting that Travis complained about and added the second paragraph to why-it's-bad.

[oli-obk]

@bors r+

superb, thanks!

[bors]

📌 Commit bc4c3b6 has been approved by oli-obk

[bors]

⌛️ Testing commit bc4c3b6 with merge 075c212...

[bors]

☀️ Test successful - checks-travis, status-appveyor
Approved by: oli-obk
Pushing 075c212 to master...