Adds support for class completions after ASI inserted class property definition

[orta]

Fixes #30536 - handles both the case where you start writing on a newline directly, and when you are continuing to write inside an identifier.

[andrewbranch]

Adding the isObjectTypeDeclaration(location) miiiight have been overly conservative, but I suggested it to err on the side of caution—the smallest reasonable net I could think of that would catch the repros provided.

[sheetalkamat]

Can you please reverse the order of checks here (that is checking isObjectTypeDeclaration(location) before line?

[orta]

Switching the if ordering changes the behavior from what it previously was, and starts to trigger behavior which I think we don't want.

This fourslash test fails:

/// <reference path="fourslash.ts" />

////function /*a*/ ;
////function* /*b*/ ;
////interface I {
////    abstract baseMethod(): Iterable<number>;
////}
////class C implements I {
////    */*c*/ ;
////    public */*d*/
////}
////const o: I = {
////    */*e*/
////};
////1 * /*f*/

verify.completions(
    { marker: ["a", "b"], exact: undefined, isNewIdentifierLocation: true },
    { marker: ["c", "d"], exact: ["baseMethod"], isNewIdentifierLocation: true },
    { marker: "e", exact: ["baseMethod"] },
    { marker: "f", includes: [{ name: "Number", sortText: completion.SortText.GlobalsOrKeywords }] },
);

With

  1) fourslash tests
       tests/cases/fourslash/completionsGeneratorFunctions.ts
         fourslash test completionsGeneratorFunctions.ts runs correctly:
     Error: At f: Expected 'isNewIdentifierLocation' to be false, got true

Which is returning an empty array of completions

[orta]

Alright I'm merging this 👍