rustdoc-json-types: Rename and document Impl::synthetic

[aDotInTheVoid]

"synthetic" impls arn't a really a term that are used, especially in user-facing documentation. It's quite confusing to users that we use this to refer to impl's of auto trait. It's lead to users beliving (and understandably so) that the JSON output doesn't contain wether an impl is auto or not. (eg, CC @zjp-CN).

We should rename this field to something more descriptive. An ititial proposal would be is_auto, but I'm not sure this is quite right, as it's the trait that's auto, not the impl. I'm open to suggestions for the field name.

While we're at it, we should document it and add tests.

[fmease]

is_auto_trait_impl?

Note that in the middle end of rustdoc, synthetic impls not only refer to (synthesized) auto trait impls but also to (synthesized) blanket impls.

I haven't checked rustdoc json but it's therefore possible that synthetic not only refers to auto trait impls but also blanket impls. If so, the field should be called is_blanket_or_auto_trait_impl.

[zjp-CN]

TIL synthetic refers to many things that are introduced by rustc:

• impl Type: GenericParamDefKind::synthetic
• synthetic GATs for each RPITIT: https://rustc-dev-guide.rust-lang.org/return-position-impl-trait-in-trait.html?highlight=synthetic#lowering-to-synthetic-associated-types
• auto traits: https://rustc-dev-guide.rust-lang.org/rustdoc-internals.html?highlight=synthetic#from-soup-to-nuts-or-an-unbroken-thread-stretches-from-those-first-cells-to-us

[fmease]

Also note that – at least in the middle end of rustdoc (but likely also in rustdoc json) –, we have yet another use of the word synthetic, namely in the context of type parameters: APITs (impl-Trait in argument position) get lowered to synthetic type parameters by rustc and rustdoc keeps up this naming scheme (rightly so). Might be worth renaming the field synthetic of type parameters to is_impl_trait (or is_apit to rules out any potential confusion with opaque types) in rustdoc json.