Add optional concrete fallback definition after overload stubs

[dyollb]

Add optional concrete fallback definition after overload stubs

Why:

Some tools (for example Griffe during re-export resolution) cannot reliably resolve symbols defined only as overload stubs. Emitting a concrete non-overload signature after the overload set preserves typing quality while ensuring symbol resolution works.

What:

1. Added a new CLI option: --print-overload-fallback.
2. Updated Printer emission flow to append one concrete fallback definition after overloaded function/method groups when the option is enabled.
3. Kept existing @typing.overload signatures unchanged for typing precision.
4. Refactored fallback processing logic to reduce duplication between function and method paths.
5. Added/updated tests to verify:
6. fallback is emitted when enabled,
7. fallback is not emitted when disabled.

[dyollb]

maybe this is a bug/issue in griffe and does not need a solution in pybind stubgen
mkdocstrings/griffe#417

[skarndev]

This does not seem to be required for using stubs for typechecking purposes. I don't seem to have an issue when running .pyi through e.g. mypy, even though more recent typechekers like zuban, ty and pyrefly need separate testing.
Maybe that should be handled on Griffe's side instead?

[dyollb]

I agree. It is even marked as bug in griffe. Closing