Attach tags and metadata to Trigger.dev v4 runs and query runs by tag through the Management API

Documented steps

1. Tag at trigger time (string or array, max 10 tags, 1-128 chars each): `await myTask.trigger(payload, { tags: ["user_123456", "org_abcdefg"] });` — also supported on `batchTrigger()` (https://trigger.dev/docs/tags).
2. Add tags from inside a run: `import { tags } from "@trigger.dev/sdk"; await tags.add("product_1234567");` and read existing ones from `ctx.run.tags`.
3. Write structured progress with the metadata API inside the run: `import { metadata } from "@trigger.dev/sdk"; metadata.set("progress", 0.5);` plus `get`, `del`, `append`, `remove`, `increment`, `decrement`, `replace` (https://trigger.dev/docs/runs/metadata).
4. Read the whole object with `metadata.current()` and force an immediate write with `metadata.flush()` instead of waiting for the batched update.
5. Propagate upward from a subtask with `metadata.parent.set(...)` / `metadata.root.append(...)`.
6. Query by tag from outside a task: `import { runs } from "@trigger.dev/sdk"; for await (const run of runs.list({ tag: "user_123456", status: ["COMPLETED"] })) { ... }` (https://trigger.dev/docs/management/runs/list).
7. Combine filters as needed — `status`, `taskIdentifier`, `createdAt.from`/`.to`/`.period`, `version`, `isTest`, `bulkAction`, `schedule`, `error` — and paginate with `size` (10-100, default 25) plus `after`/`before` cursors.
8. Fetch one run's metadata after the fact with `const run = await runs.retrieve("run_1234"); run.metadata;` authenticating with the environment's secret key.