Refactor ContentType + improve content type matching

[czechboy0]

Motivation

The ContentType type had two issues:

• the enum where all cases had the same associated string value was not a good representation of the data
• the parsing was too strict and didn't match e.g. foo/bar+json as a JSON content type

Modifications

Refactored ContentType to split the raw value storage from the Category (new term, ideas of a better name are welcome) of content types (json, text, or binary).

Also, improved the mapping to match https://json-schema.org/draft/2020-12/json-schema-core.html#section-4.2

Result

Now the type is easier to work with, and parsing correctly recognizes foo/bar+json as JSON.

Test Plan

Added unit tests.

[glbrntt]

Looks good but could you add a comment to the binary case before merging?

[czechboy0]

@glbrntt, actually I made a few more changes in 91828ac, could you take a look if that still has your approval? It was aimed at avoiding callers accidentally using case sensitive comparisons of content types, plus it allows access to the parsed type and subtype components.