test(rsc): make e2e file editor async and space out writes

[hi-ogawa]

• split out from test(rsc): fix e2e flakes in use-cache-persistent and isolated fixture setup #1487, originally found in fix(rsc): resolve facade client entry with strictExecutionOrder #1480 by @james-elicx
• example failure on main: dev-no-ssr › client hmr on windows, stuck waiting for Client Counter: 1 after the reset on every retry

Back-to-back file edits in e2e could be silently dropped. Vite's file watcher ignores a change that arrives within 50ms of the previous one for the same path, so when the HMR round trip was fast, a test's reset landed inside that window and the dev server never saw it. The dev server then kept the edited content, and the test timed out waiting for the original text, as in the failure linked above.

The test file editor now waits at least 100ms between writes to the same file. Waiting needs to be async, so editing, resetting, and resaving now return promises and every test awaits them. The delay only happens when two writes to one file come back to back.

await editor.edit((s) => s.replace('Client Counter', 'Client [edit] Counter'))
await expect(page.getByRole('button', { name: 'Client [edit] Counter: 1' })).toBeVisible()
await editor.reset()

A missing await would not fail loudly, because the write would just happen a moment later. The e2e directory has no lint step to catch that, so every call site was updated in one pass.