Concurrency Limits Per Context / Provider

Workaround

Create dedicated worker pools with a concurrency of 1 for stacks that need to be throttled against a specific provider. This works but is operationally expensive and doesn't scale well for large numbers of stacks.

Problem

Users can hit rate limits on remote APIs when multiple stacks using the same context run concurrently. They want to be able to set a concurrency limit at the context level - e.g. "only 1 run at a time may use this context" - without having to create and manage separate worker pools purely for throttling purposes.