[Saiop68]

Adds BetterQueue, a distributed rate limiter and job queue for Node.js.

The queue state lives in Redis rather than in the process, which changes three practical things:

• A limiter is a thin handle. Creating 1000 limiters costs roughly what 10 do, because the per-limiter state is not held in the process.
• Jobs survive a crash. A job held by a worker that dies is re-queued and picked up by another process instead of being lost.
• Priority ordering is global. Ordering is computed server-side in Redis, so a high-priority job outranks a lower-priority one in any process, not just within one limiter.

It also supports scheduleAt() for work that needs to run later, which Bottleneck does not.

The repository includes a benchmark that runs both libraries in separate processes against the same Redis. With 1000 limiters and 51,000 jobs:

Metric	Bottleneck 2.19.5	BetterQueue
RSS after the run	415 MB	222 MB
Heap after forced GC	14 MB	11 MB
Throughput	9,957 jobs/s	11,456 jobs/s

MIT licensed, written in TypeScript, 37 tests with CI running on Node 18/20/22.