task: make max_concurrent_execution global to a client instance

[fatih-acar]

Component

Python SDK

Task Description

#476 #483
Highlighted an issue with using a shared semaphore for batches.
When using batches within batches, the process would hang forever.

Root cause is unknown yet, but in order to have a consistent behavior with the InfrahubClientSync where batches leverage ThreadPools with a local max workers values, we removed the shared semaphore in #601

We need another approach to the Sync client threadpool solution and fix the shared semaphore for async in order to have max_concurrent_execution global to a client instance.