make launch return results from all ranks

Summary:
X-link: https://github.com/facebookresearch/mobile-vision/pull/86

Pull Request resolved: https://github.com/facebookresearch/d2go/pull/300

Similar to `elastic_launch`, make `launch` return results for all ranks, this is beneficial for cases where each rank having different return value`.

There'll be some test failing for V1, it'll help finding related callsites.

Reviewed By: tglik

Differential Revision: D37016935

fbshipit-source-id: eb989342c0ddd642c8369b5dddfa8aac8cb38917

d2go/distributed.py

@@ -68,7 +68,7 @@ def launch(
     timeout: timedelta = DEFAULT_TIMEOUT,
     args: Tuple[Any, ...] = (),
     kwargs: Dict[str, Any] = None,
-):
+) -> Dict[int, Any]:
     """
     D2Go's specialized launch method, it does a few more things on top of mcv's launch:
         - Automatically convert GPU to CPU if CUDA is not available.