fw-03 step 03 — Scheduling, then rebuild it as signals

Goal

Batch the effects, then discover how much of your Vue implementation was essential versus Vue-specific.

Tasks

1. Scheduler. A deduplicated queue flushed on a microtask. Reproduce the un-batched case first: 1,000 writes in a loop, measured.
2. Explain why a microtask is the right boundary, referencing bi-11.
3. Now build signals.js from scratch — signal, computed, effect, batch. Do not refactor the Vue code into it. Copy the spec, adapt the API surface, and get it green.
4. Compare the two implementations. What was essential? What was Vue-specific? The answer is the finding.
5. Fill your version of the four-model comparison table (React / Vue / signals / Redux) across: how it knows what changed, unit of update, what cost scales with, whether "state at time T" exists, and what tooling each demands.

Done when

• Batching measured before and after
• signals.js written independently and passing an adapted spec
• Comparison table filled in your own words
• Notebook entry: the Vue scheduler