Feature request: add a buffered “throttle” utility that queues excess calls

[bytesbay]

I’m using TanStack Pacer to rate-limit async calls at 100 requests/sec. The built-in AsyncRateLimiter rejects calls beyond the cap, and queue/AsyncQueuer need manual config to approximate a “p-throttle”–style behavior (limit+buffer).

Would you consider adding a first-class, buffered throttle utility (à la [p-throttle](https://github.com/sindresorhus/p-throttle)) that:

1. Restricts to a configurable number of executions per window (e.g. 100/sec)
2. Automatically queues any overflow for the next window, without dropping or rejecting calls

This would simplify common API-rate-limit patterns and avoid the need to compose multiple primitives. Thanks!

[KevinVandy]

As you say, it sounds like this functionality can be made by composition of a rate limiter and queuer. On rejection, add items to an overflow queue? I guess we could make some first class composites like this, but it's probably more flexible for us to just document how to do it.