Define Trigger.dev v4 queues with concurrencyLimit and isolate per-tenant throughput with concurrencyKey

Documented steps

1. Create a reusable queue: `import { queue } from "@trigger.dev/sdk"; export const myQueue = queue({ name: "my-queue", concurrencyLimit: 1 });` then assign it to tasks so they share one limit (https://trigger.dev/docs/queue-concurrency).
2. Or configure inline on a single task: `task({ id: "one-at-a-time", queue: { concurrencyLimit: 1 }, run: ... })`.
3. For per-tenant isolation on a shared queue, pass `concurrencyKey` at trigger time: `await generatePullRequest.trigger(data, { queue: "free-users", concurrencyKey: data.userId });`
4. Override the queue per run at trigger time with `{ queue: "paid-users" }`.
5. Account for environment-level concurrency: the environment has a base limit plus a burstable limit (default burst factor 2.0x base); an individual queue is capped at the base limit, not the burst capacity.
6. Give subtasks their own queue config explicitly — subtasks do not inherit the parent task's queue.
7. Watch queue depth and concurrency in the dashboard when tuning limits.