Add Poller<T> for long-running operations (LROs)

[heaths]

Resolves #2756 and resolves #2482

[heaths]

I still need to implement the retry options (will likely mimic our RetryOptions for the RetryPolicy) but I wanted to get feedback on the DX here. Since implementing IntoFuture for #2757 is on the backlog and would return the StatusMonitor::Output, I took a page from other uses of wait() in both std::process::Command::wait() and tokio::process::Command::wait() and return a future that will just pump the Stream implementation until terminal state and return the last status monitor.

This may be temporary until we implement #2757, though maybe still valid. This at least makes tests, examples, and doc examples easier to understand; otherwise, we'd have to add loops for everything and just ignore every intermediate status monitor.

[github-actions]

API Change Check

APIView identified API level changes in this PR and created the following API reviews

azure_core
azure_security_keyvault_certificates
typespec_client_core

[magodo]

@heaths I have a working LRO (for ARM mgmt plane API) implementation, which naively port the azure-sdk-for-go to Rust: https://github.com/magodo/azure/tree/main/src/client, if you are interested in.

[heaths]

@magodo I'm unable to access it. But I will say that, while I looked at Go's and other languages (I worked on the Azure SDK for .NET team previously), it's not a pattern I want for the Rust SDK. The pollers do too much at runtime that, because we are strictly TypeSpec-based, we can do at compile time. The current design here not only has a consistent code flow with our Pager, but makes it easy to also get the final object in a subsequent IntoFuture implementation.

[LarryOsterman]

The extension methods for the poller are ugly but necessary.

[heaths]

@microsoft-github-policy-service rerun

[heaths]

/check-enforcer evaluate

[heaths]

@microsoft-github-policy-service rerun

[LarryOsterman]

/check-enforcer evaluate

[heaths]

@microsoft-github-policy-service rerun

[heaths]

/check-enforcer evaluate