document trait bounds and candidate preference behavior

[lcnr]

cc #3

documents http://github.com/rust-lang/rust/pull/120752

would be nice for somebody to quickly skim over this to detect

• things which still need to be defined
• should be moved

how to talk about X

how to talk about relating higher ranked types. "instantiating the for of the subtype with inference variables and the for of the supertype with placeholders before relating them as normal" seems quite cumbersome. This is where using judgements/actual code is actually ncier than text.

concepts which I don't know how to avoid

rigid types: a type which cannot be normalized in teh current context, whose equality is implemented by equating the type constructor and then recursively their corresponding generic arguments

• why useful: we normalize aliases until we encounter a rigid type constructor? either a rigid alias or an a non-alias

[joshtriplett]

A couple of nits and a question, but the overall structure and organization of this seems potentially reasonable.

[lcnr]

🤷 idk how to make progress here, would be nice if someone could look over this again.