Parse doc comment markdown only once

[Alexendoo]

doc_link_code and doc_paragraphs_missing_punctuation are now state machines that take in the events from check_doc's markdown parser

changelog: none

[rustbot]

Some changes occurred in clippy_lints/src/doc

cc @notriddle

[rustbot]

r? @blyxyas

rustbot has assigned @blyxyas.
They will have a look at your PR within the next two weeks and either review your PR or reassign to another reviewer.

Use r? to explicitly pick a reviewer

[notriddle]

There's a bug in this new version that isn't in the current version. Unfortunately, there's no test case for them in the code, but here's a couple of test cases that catch it (they pass in the main branch, but fail in this PR's branch):

/// `first`[`second`](https://example.com)[third](https://example.com)
//~^ ERROR: adjacent
pub struct FalseNegative;
/// `first`[](https://example.com)
pub struct FalsePositive;

[Alexendoo]

Thanks, it now verifies that links contain a single code and nothing, which catches a FP on:

`first`[`second` not code](x)

Separate to this PR but I think it'd be worth moving out of nursery unless there's a known issue I'm missing

[Alexendoo]

As an aside what does the arst at the end of the test file lines mean? As a brit I cannot help but read it as "arsed"

[notriddle]

I use the Colemak layout, and arst just happens to be the left hand home row keys.

[blyxyas]

Benchmarks results: Not noticeable (-0.15% on tokio)