Complete generator support for return value

[atifaziz]

This PR completes the implementation of generator support by adding and testing the ability to retrieve the return value from a generator function.

Summary of changes:

• A new property Return is added to IGeneratorIterator<,,> that behaves like IEnumerator<>.Current.
• Send of IGeneratorIterator<,,> is changed to match semantics of MoveNext. As an aside, I would even propose to rename Send to MoveNext on the C# side since overloading the same name is more natural given that next in the Python is the same as send(None).
• Send implementation of GeneratorIterator<,,> is changed to:
  • match semantics of MoveNext. The latter would return false whereas the former would throw ArgumentOutOfRangeException (with the wrong parameter for message) when the generator was exhausted.
  • retrieve the StopIteration.value.
• MoveNext implementation of GeneratorIterator<,,> is changed to simply delegate to the internal Send implementation with PyObject.None for value. This matches the Python semantics:

  New generator method: send(value) Calling send(None) is exactly equivalent to calling a generator’s next() method.

• A new exception called PythonStopIterationException (that subclasses PythonRuntimeException) is introduced and which captures the value of StopIteration when it is raised so that it can be retrieved later. I don't think this is a great design on the .NET side, but this can be revisited later since the project is still a prototype. For now, it helps to complete generator support. Unfortunately, the implementation is also complicated by what I think is a regression in PyErr_Fetch that's been introduced with 3.12 (see comment in code).

[tonybaloney]

Please can you extend this test so that it verifies that the generator returned True, or change the generator to return something more exotic incase we accidentally return a truthy value anyway

https://github.com/tonybaloney/CSnakes/blob/main/src/Integration.Tests/GeneratorTests.cs#L14

[atifaziz]

Please can you extend this test so that it verifies that the generator returned True, …

I'm not sure what's missing since the test was already extended.

[aaronpowell]

Looks good to me pending @tonybaloney's desires on testing

[tonybaloney]

Sorry I didn't see the additional test change. LGTM