rustc: Make the trait_map of TyCtxt private

[alexcrichton]

This map is calculated in resolve, but we want to be sure to track it for
incremental compliation. Hide it behind a query to get more refactorings later.

cc #44137

[rust-highfive]

r? @arielb1

(rust_highfive has picked a reviewer for you, use r? to override)

[arielb1]

I'm worried about the proliferation of providers, but I think making trait_map public to rustc::ty would be worse.

[arielb1]

I don't like the proliferation of provide methods, but I think making trait_map public in all of rustc::ty would be worse.

[arielb1]

Basically LGTM, but there's a travis blip and I suspect the primary use-case of TraitMap in

rust/src/librustc_typeck/check/method/probe.rs

Lines 662 to 678 in faf477a

	fnassemble_extension_candidates_for_traits_in_scope(&mutself,
	expr_id: ast::NodeId)
	-> Result<(),MethodError<'tcx>>{
	letmut duplicates = FxHashSet();
	let opt_applicable_traits = self.tcx.trait_map.get(&expr_id);
	ifletSome(applicable_traits) = opt_applicable_traits {
	for trait_candidate in applicable_traits {
	let trait_did = trait_candidate.def_id;
	if duplicates.insert(trait_did){
	let import_id = trait_candidate.import_id;
	let result = self.assemble_extension_candidates_for_trait(import_id, trait_did);
	result?;
	}
	}
	}
	Ok(())
	}

hasn't been updated.

[alexcrichton]

Ok I think I handled that case (still waiting on a local compile)

I also switched this over to being keyed on a HirId at @nikomatsakis's suggestion

[alexcrichton]

Ok I've also pushed up a commit which hides the export_map in a similar fashion.

re-r? @arielb1

also cc @nikomatsakis

[arielb1]

@bors r+

[bors]

📌 Commit 942c8dc has been approved by arielb1