Summary

Closes #2086

Deferred from #2031's review (Greptile 5/5, one open note): collectLocalAccessors/collect_local_accessors's same-file ES6 get/set accessor registry keyed purely by ClassName.propName -> {get, set}, with no static/instance distinction. this inside an instance method never refers to the class/constructor object (where static members live) — only this inside a static method does. If a class declared only a static get X() (no instance accessor of that name), a bare this.X read inside an instance method matched against the static accessor and produced a false-positive calls edge (and the mirror case — an instance-only accessor read from a static method — had the same gap in reverse):

export class Config {
  static get version(): string { return Config._v; }
  static _v = '1.0';

  // this refers to the Config INSTANCE here — this.version does not invoke
  // the static getter at runtime — yet codegraph reported
  // Config.describe -> Config.version.
  describe(): string {
    return this.version;
  }
}

Changes

• src/extractors/javascript.ts / crates/codegraph-core/src/extractors/javascript.rs: LocalAccessorInfo gains staticGet/staticSet fields alongside the existing get/set (now unambiguously instance-only). collectLocalAccessors/collect_local_accessors routes each declared accessor into the correct bucket via a new isStaticMethodDefinition/is_static_method_definition check.
• A new isEnclosingMethodStatic/is_enclosing_method_static walks up from a this.prop read to the nearest enclosing method_definition and reports whether it's static — that determines which bucket (static* vs instance) the read must match against. Only meaningful once findParentClassForThisBinding (follow-up: call/property-read attribution doesn't respect this-binding scope boundaries #2085) has already confirmed no this-rebinding boundary sits in between; an arrow function is transparent to both walks.
• The varName.prop (non-this) branch is untouched: it already only needs the instance bucket, since typeMap never records a variable holding the class object itself — so unifying get/set to mean "instance" for free fixes that branch's equivalent gap too (an instance-typed variable reading a static-only accessor).
• Tests added on both sides for all four combinations: instance-context reading a static-only accessor (no edge), static-context reading an instance-only accessor (no edge), and confirming the pre-existing static-to-static and instance-to-instance cases still pass.

Known narrow gap left out of scope (matches the issue's stated boundary): a class field initializer or static { } block isn't inside any method_definition, so isEnclosingMethodStatic defaults to instance-context there — a static field initializer's own this.prop read could still misclassify. Not handled here; flagged in code comments for anyone extending this further.

Test plan

• cargo test -p codegraph-core --lib — 822 passed (including 3 new tests)
• npx vitest run tests/parsers/javascript.test.ts — 311 passed (including 3 new tests)
• npm test — full suite, 4545 passed
• npx tsc --noEmit clean
• cargo build -p codegraph-core --lib clean
• Manual native-vs-wasm parity check via codegraph build --engine native|wasm + codegraph query against the issue's repro (both engines agree: Config.version has no callers, Widget.value is called only by the instance method, not the static one)
• codegraph diff-impact --staged -T reviewed