Batch-trigger many runs of one Trigger.dev v4 task efficiently and handle batch limits and rate limiting

Documented steps

1. Import `tasks` from `@trigger.dev/sdk` plus a type-only import of the task (https://trigger.dev/docs/triggering).
2. Call `const batchHandle = await tasks.batchTrigger<typeof emailSequence>("email-sequence", users.map(u => ({ payload: { to: u.email } })));` — each item is an object with a `payload` field.
3. Set per-item options with an `options` field on the item (e.g. `{ payload, options: { delay: "1h" } }`), or shared options via the third argument.
4. Read `batchHandle.batchId`, then call `batch.retrieve(batchHandle.id)` to get the batch with its runs — in v4 the handle no longer exposes runs directly (https://trigger.dev/docs/upgrade-to-v4).
5. For large or unbounded datasets pass an `AsyncIterable`/async generator or a `ReadableStream` instead of materializing one giant array.
6. Respect the documented limits: 1,000 items per batchTrigger call on SDK 4.3.1+ (500 on earlier v4 SDKs), and 3MB per item on SDK 4.3.1+ (https://trigger.dev/docs/limits).
7. Catch `BatchTriggerError` and inspect `isRateLimited`, `retryAfterMs`, `phase` and `batchId` to back off correctly.