percent_encoding: normalise_percent_encoded and strict_percent_encoded

[WhyNotHugo]

As part of an unrelated WebDAV specific project, I've written two utility functions closely tied to percent encoding:

• normalise_percent_encoded(): Decodes percent-encoded strings while keeping RFC 3986 reserved characters encoded (with uppercase normalization).
• strict_percent_encoded(): Encodes non-reserved characters for safe URI paths.

These are actually not webdav-specific at all, and re-usable in other HTTP-related contexts.

Do you consider these a suitable contribution for the percent_encoding crate? If so, I can draft a PR incorporating both of these. The implementation for these is quite non-trivial, and they have been heavily optimised to minimise allocations and heavily tested.