fix(resolution): a store-action binding is not a local shadow

[bompus]

Fixes #1762.

#1759 made isLocallyBoundJsName treat a Zustand-style store binding as a local rebinding, so bare calls to store actions stopped resolving. main at cd4e65b currently fails its own __tests__/object-literal-methods.test.ts with expected [] to include 'loginFlow', plus two tests in __tests__/ui-steps-api.test.ts.

The change

isLocallyBoundJsName already carves out require(...) / import(...) on the right-hand side, because an imported name is a node elsewhere rather than a shadow. This extends that same carve-out to store bindings, which are the same situation — the action is a node in the store file, so the bare call that follows means it:

const { fetchUser } = useStore.getState();   // accessor shapes matchStoreAccessorChain already knows
const { fetchUser } = get();                 // inside the factory
const setZipUri = useStore((s) => s.setZipUri);  // selector reading the same name

A selector returning a different name is a genuine rebinding and still shadows.

The selector pattern is deliberately narrow: one named parameter, and a plain member read of the same name ((?!\s*\()). const now = options.now || (() => Date.now()) matches neither clause — no parameter, and .now() is a call — so it stays the local binding this predicate exists to find. bare-call-no-method covers that case and stays green.

Verification

Full suite, same machine and same build, pristine cd4e65b vs this branch:

	pristine	this branch
Test files	13 failed	7 failed
Unique failing tests	29	24

object-literal-methods and both ui-steps-api tests go green, and no test that passed before fails after. The remaining 24 are pre-existing on this machine (Windows, native kernel not built) and identical across both runs.

[bompus]

Duplicate of #1760, which fixes the same regression. Closing; verification data moved to #1760.