Assorted find-all-references bugs

I made a list of things I discovered while working on #13760.

• Some tests don't have isWriteAccess when they should

  • findAllRefsParameterPropertyDeclaration1 (/2, /3): this.privateParam += 10 should be a write access
  • referencesForNumericLiteralPropertyNames: These should be write accesses if normally-named properties are
  • referencesForStringLiteralPropertyNames (/4): Same as above
  • referencesForClassParameter: All should be write accesses
  • localGetReferences (e.g. fooCls.clsSVar++)
  • getOccurrencesIsDefinitionOfComputedProperty: definition should be a write access
  • referenceInParameterPropertyDeclaration
  • remoteGetReferences

• Some tests have multiple groups with identical definitions.

  • findAllRefsForObjectLiteralProperties
  • findAllRefsOnDefinition: r1
  • findAllRefsWithLeadingUnderscoreNames1 (/2, /3, /4)
  • findAllRefsWithShorthandPropertyAssignment: r4
  • findAllRefsWithShorthandPropertyAssignment2: r3
  • referencesForInheritedProperties: r3
  • referencesForInheritedProperties2: r3
  • referencesForInheritedProperties4: r1
  • referencesForObjectLiteralProperties
  • referencesForPropertiesOfGenericTypes
  • ambientShorthandFindAllRefs
  • getReferencesAtPosition
  • getOccurrencesIsDefinitionOfStringNamedProperty
  • referencesForClassParameter
  • referencesForClassLocal
  • referencesBloomFilters (/2)
  • getOccurrencesIsDefinitionOfNumberNamedProperty
  • cancellationWhenFindingAllRefsOnDefinition
  • referencesForClassMembers: m2

• Some tests have a different ordering of merged declarations depending on where you start.
  (Maybe this is fine.)

  • findAllRefsForDefaultExport03
  • findAllRefsInheritedProperties1
  • findAllRefsInheritedProperties3 (see r6)
  • referencesBloomFilters (/2)

• Other

  • findAllRefsForComputedProperties: I[["prop1"]] should be just I["prop1"].
  • findAllRefsForMappedType:
    It's iffy that (property) a: string only shows up when you're directly using it, and otherwise is grouped under (property) T.a: number.
  • findAllRefsObjectBindingElementPropertyName06: r4 could be just one definition
  • findAllRefsObjectBindingElementPropertyName07:
    The first a shouldn't be a write access (or a definition): it's just the property name we're destructuring from. p is the definition.
  • findAllRefsOnImportAliases2: In import Class as C2, Class doesn't really deserve "isDefinition"
  • findAllRefsThisKeyword:
    Wierd that "f0" and "f1" have different definitions. Same for "propDef" and "propUse".
  • findAllRefsForObjectSpread:
    A2.a: number should be A2.a?: number
  • referencesForContextuallyTypedUnionProperties (/2):
    Strange how most of them are considered A.common.
  • referencesForEnums:
    If value1 is a write access, value2 should be too
  • referencesForMergedDeclarations: two of the definitions do not contain "interface Foo", why?
  • renameDefaultImportDifferentName:
    "constructor C(): B" is iffy, since 'C' and 'B' are the same thing
  • getOccurrencesIsDefinitionOfComputedProperty: Should work for r1 and r2
  • referencesForClassMembersExtendingGenericClass:
    Wierd that for m1, it shows as Base<T>.method(): void

These are all fixed now. 🎊