Improve go-to-definition and implement go-to-type-definition

[ahejlsberg]

This PR implements

• Go-to-def on override keyword selects the overridden method declaration.
• Go-to-def on a modifier or keyword that starts a declaration works the same as go-to-def on the declaration name.
• Go-to-def on label in break or continue statement selects the label definition.
• Go-to-def on case or default keyword selects the containing switch statement.
• Go-to-def on return, yield or await keyword selects the containing function declaration.
• Go-to-def on a call expression selects the the signature chosen by overload resolution.
• Go-to-def on a property in an object literal with a contextual type selects the declaration in the contextual type.
• Go-to-def on a property with no explicit declaration selects the index signature(s) for the property.
• Go-to-type-definition on a variable or property selects the associated type(s).
• Go-to-type-definition on a variable with a generic type such as T[] selects both declarations of T and Array.

The PR also moves getErrorRangeForArrowFunction and GetErrorRangeForNode from binder.go to scanner.go (where other similar functions reside).

[Copilot]

Pull Request Overview

This PR enhances the go-to-definition functionality and implements go-to-type-definition for the TypeScript language server. It adds support for navigating to definitions from various language constructs including override keywords, labels, control flow statements, and call expressions, while also introducing type definition capabilities.

Key changes:

• Extended go-to-definition to handle keywords (override, case, default, return, yield, await) and jump statements
• Added comprehensive go-to-type-definition support with type argument handling
• Moved error range calculation utilities from binder to scanner for better organization

Reviewed Changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
internal/scanner/scanner.go	Moved GetErrorRangeForNode function from binder and simplified GetTokenPosOfNode
internal/lsp/server.go	Added type definition endpoint handling and server capability registration
internal/ls/utilities.go	Updated range creation to use token position and fixed label handling return type
internal/ls/findallreferences.go	Updated label reference functions to work with generic nodes instead of identifiers
internal/ls/definition.go	Major expansion with keyword-based definitions, call resolution, and type definition implementation
internal/checker/utilities.go	Updated diagnostic creation to use scanner's error range function
internal/checker/services.go	Added contextual declarations and type argument extraction utilities
internal/checker/exports.go	Exported new checker methods for index signatures and symbol resolution
internal/checker/checker.go	Implemented index signature location resolution
internal/binder/binder.go	Removed GetErrorRangeForNode function (moved to scanner)
internal/ast/ast.go	Added IsSwitchStatement helper function

[RyanCavanaugh]

Suggested change

	"Omit": {},
	"Omit": {},
	"Exclude": {},

[ahejlsberg]

The types in this list generally operate on object types for which we record source locations. Exclude and Extract are almost always used with (unions of) literal types, and we don't record source locations for those.

[ahejlsberg]

However, we should include NonNullable<T>.