[braket] Expose Amazon Braket quantum task ARN from `cudaq.sample_async` and `cudaq.observe_async`

Required prerequisites

• Search the issue tracker to check if your feature has already been mentioned or rejected in other issues.

Describe the feature

The CUDA-Q doc for algorithmic primitives (sec 12.1 cudaq::sample [12]) specifies that "For remote QPU systems with long queue times, the async_sample_result type encodes job ID information and can be persisted to file and loaded from file at a later time...". While the async_results for a job submission returns a future-like object, it does not expose the underlying Amazon Braket job ARN for retrieval in a separate process. Without access to the Braket job ARN, a separate process or poll cannot retrieve results from the original CUDA-Q process via an event bus such as Amazon EventBridge. cudaq.sample_async(...) / cudaq.observe_async(...) should allow users to access the underlying Braket quantum task ARN (or other job id), so results can be retrieved later by a different process, or expose an API for retrieving the results of a QPU job submission via a job id.

Start with the algorithmic-primitives documentation for async_sample_result and trace the cudaq.sample_async and cudaq.observe_async entry points into the Amazon Braket integration. Determine how a Braket quantum task ARN or equivalent job ID could be exposed or used for later retrieval, and define completion as supporting retrieval from a separate process.