refactor(headers): Improve docs, fix nits, make formatting faster

[pyfisch]

src/header/parsing.rs now uses unsafe get_unchecked() again, I don't
know why it was removed.

[reem]

LLVM almost certainly elides the bounds check here anyway. I'd personally rather have less unsafe. I doubt this has a noticeable impact on perf, but I'd be thrilled to be proven wrong.

[seanmonstar]

^{src/header/parsing.rs, line 12 [r1] (raw file):}
Either way seems fine to me. Or rather, if the "almost certainly" is in fact "absolutely always", then cool. Otherwise, we've done the check above, and a comment reminding anyone about the invariant, so the unsafe looks safe to me as well.

---

---

Comments from the review on Reviewable.io

[seanmonstar]

Reviewed files:

• src/header/common/accept.rs @ r1
• src/header/common/access_control_allow_headers.rs @ r1
• src/header/common/access_control_allow_methods.rs @ r1
• src/header/common/access_control_max_age.rs @ r1
• src/header/common/access_control_request_headers.rs @ r1
• src/header/common/access_control_request_method.rs @ r1
• src/header/common/allow.rs @ r1
• src/header/common/authorization.rs @ r1
• src/header/common/cache_control.rs @ r1
• src/header/common/connection.rs @ r1
• src/header/common/content_type.rs @ r1
• src/header/common/cookie.rs @ r1
• src/header/common/etag.rs @ r1
• src/header/common/host.rs @ r1
• src/header/common/if_range.rs @ r1
• src/header/common/last_modified.rs @ r1
• src/header/common/mod.rs @ r1
• src/header/common/pragma.rs @ r1
• src/header/common/set_cookie.rs @ r1
• src/header/internals/item.rs @ r1
• src/header/mod.rs @ r1
• src/header/parsing.rs @ r1
• src/header/shared/charset.rs @ r1
• src/header/shared/encoding.rs @ r1
• src/header/shared/entity.rs @ r1
• src/header/shared/language.rs @ r1

---

Comments from the review on Reviewable.io

[seanmonstar]

Hrm, seems reviewable is noisier than I was hoping. Oh well, don't worry about it. Just playing with toys, and this was a larger PR, so seemed opportune.