Suggest completions for type arguments of expressions

[Matt Kantor (mkantor)]

Completions are now suggested within type arguments of the following expression types (based on type parameter constraints):

• function calls (f<…>())
• new expressions (new Foo<…>())
• tagged templates (tag<…>`blah`)
• JSX elements (<Component<…>/>)
• decorators (@decorator<…> class {})
• instantiation expressions (f<…>)

Fixes#61751.

Limitations

• Completions for a property values (when the cursor is after a : in a type argument literal) still don't happen, so No completions from tsserver for property values in generic type arguments #56299 should remain open. I might take a stab at this, but it'll be a separate effort from the present changeset.EDIT: I took a stab at this and have something that works, but it's stacked atop this branch so I'll see where this pull request goes before opening another one.
• Completions don't work within string literal types. I started looking into this, but ran into problems (that's considered a "known invalid completion location" according to getCompletionData in completions.ts). This needs fixing to fully address the specific examples in No string completions in type arguments where constrained to string literal union #52898 and No IntelliSense for generic literal types #34771. EDIT: I also have a (stacked) fix for this.

This comment may also be useful reading.

[TypeScript Bot (typescript-bot)]

Looks like you're introducing a change to the public API surface area. If this includes breaking changes, please document them on our wiki's API Breaking Changes page.

Also, please make sure Daniel Rosenwasser (@DanielRosenwasser) and Ryan Cavanaugh (@RyanCavanaugh) are aware of the changes, just as a heads up.

[Matt Kantor (mkantor)]

After some consideration I removed 53d3f29 from this branch, since it's pretty tangential. If there's appetite for it I could submit another pull request.

This means @typescript-bot's comment is no longer relevant.

[Matt Kantor (mkantor)]

I got nervous about keeping all of the stacked improvements to myself, so opened #62170 which includes them (along with the changes from this pull request). I'll probably close this one under the assumption that these enhancements are all desirable (if only loosely coupled to one another).

[Matt Kantor (mkantor)]

As mentioned above, see #62170 instead. If you'd prefer that I instead submit each of those commits as a separate pull request (as was my original intention), please let me know.