fix(useMap): skip rerenders for unchanged values

[sridharkalaibala]

What is the current behavior, and the steps to reproduce the issue?

With const map = useMap([['count', 0]]), calling map.set('count', 0) schedules a render on every call even though the entry has not changed.

What is the expected behavior?

An unchanged entry should not schedule a render. Adding a key or changing its value should still render, and Map.set should retain its normal return value for chaining.

How does this PR fix the problem?

Check key presence and compare its previous value with Object.is before invoking the underlying Map.set. Only schedule a render when the entry changes. The presence check means a missing key with an undefined value is still inserted and triggers a render. NaN compares equal; 0 and -0 values remain distinct, as with React state.

The hook JSDoc documents the comparison and its object-reference consequence: mutating an object and setting the same reference no longer forces a render; callers should replace the value. No changes to clear/delete or exported types.

Fixes #1655.

Validation: nine added regression/control cases cover unchanged primitives, NaN, undefined, shared object identity, changed values, signed zero, distinct objects, and insertion of an undefined entry. Original implementation: four fail and 19 pass across the useMap DOM/SSR tests. With the fix, the complete suite passes: 541 tests in 117 files, including SSR. yarn fmt, yarn lint:fix, final yarn lint/yarn fmt:check, and yarn build pass. Formatting's unrelated checkout line-ending changes were restored; only the two useMap files are included. Node 24.19, pinned Yarn 4.17.1, immutable install; no dependency/lockfile changes. Prepared with AI assistance.

Checklist

• Have you read the contribution guideline?
• Have you added an explanation of what your changes do and why you'd like us to include them?
• Is there an existing issue for this PR? useMap's set() should not re-render when setting a key to the same value it's currently set to #1655
• Have the files been linted and formatted?
• Have the docs been updated to match the changes in the PR?
• Have the tests been updated to match the changes in the PR?
• Have you run the tests locally to confirm they pass?