feat(ui-automation): Add rs/1 runtime automation parity - #416

The type-text success snapshot test was deleted and replaced with an error case only, leaving the success-path contract entirely unvalidated; no type-text--success fixture exists anywhere in the repo.

In gesture.test.ts, the Command Generation tests define inline trackingExecutor closures with a command[1] !== 'describe-ui' guard rather than using the createTrackingExecutor helper that is already imported from ui-action-test-helpers.ts. The shared helper centralizes the describe-ui JSON shape ({ elements: [createNode()] }) and call-capture format, so the inline copies drift from the canonical mock and have to re-implement the post-action describe-ui carve-out by hand. Tests still pass because captureRuntimeSnapshotAfterActionSafely swallows JSON parse errors in a try/catch, but this violates the skill's "Use existing mock executor helpers" guardrail and risks future divergence as the snapshot contract evolves.

Also found at:

• src/mcp/tools/ui-automation/__tests__/gesture.test.ts:233-241

Replace the inline createBootedScreenshotMockExecutor factory with the existing createCommandMatchingMockExecutor helper from mock-executors.ts, which already handles multi-command dispatch by pattern match.

Replace createFailingExecutor(error) with createMockExecutor({ success: false, error }) from src/test-utils/mock-executors.ts to follow the existing mock executor helpers guardrail.