More robust "KnownNames" resolution

[davidsemakula]

Description

Improves KnownName resolution by:

• Matching against rustc lang items (by DefId) when possible (should produce the most precise match, and also likely the least expensive as a lang items map is available via a rustc query)
• Accounting for Namespace when matching DefPath segments
• Making sure all path segments are consumed when matching a KnownName to the full DefPath (See Assertion failure in checker/src/call_visitor.rs:2038 #26 (comment) for details)
• Matching against the DefPathData variant instead of the integer disambiguator for DefPath segments that don't include a name (e.g. impl, extern/"foreign module" definitions e.t.c)

Fixes #26

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• API change with a documentation update
• Additional test coverage
• Code cleanup or just keeping up with the latest Rustc nightly

How Has This Been Tested?

Added a test case as described at #26 (comment)

Also ran ./validate.sh

Checklist:

• Fork the repo and create your branch from main.
• If you've added code that should be tested, add tests.
• If you've changed APIs, update the documentation.
• Ensure the test suite passes.
• Make sure your code lints.

[davidsemakula]

@hermanventer I left a few minor TODO comments for myself that I don't think stop this from landing, I plan to pick them up as separate PR(s) in the next few days or early next week.

[hermanventer]

It seems so and, in general, the Rust compiler folks are prone to renaming functions and moving functions between namespaces so the mapping provided by KnownNames continually breaks. I make time to keep the tests mostly running, but finding the time to keep KnownNames current in every respect has been difficult.

[hermanventer]

It may be worthwhile to write a test case that contains, one way or another, code the generates DefIds for each of the cases in KnownNames. Fixing such a test case when the compiler churns will be much more efficient than analyzing other test cases that break in weird ways because KnownNames is invalidated by a compiler change.

[davidsemakula]

I think it's a good idea to keep it and the TODO/comment around till we hunt down it's replacement :)

[davidsemakula]

It may be worthwhile to write a test case that contains, one way or another, code the generates DefIds for each of the cases in KnownNames.

Makes sense

[davidsemakula]

Will work on this and ping you :)

[hermanventer]

Looks good.