Fix typing error of torch/optim/lr_scheduler.pyi (#41775)

Summary:
* add `_LRScheduler.get_last_lr` type stub.
* remove `CosineAnnealingWarmRestarts.step` because its signature is same with `_LRScheduler`'s.

Pull Request resolved: pytorch/pytorch#41775

Reviewed By: izdeby

Differential Revision: D22649350

Pulled By: vincentqb

fbshipit-source-id: 5355dd062a5af437f4fc153244dda793a2382e7e

torch/optim/lr_scheduler.pyi

@@ -5,6 +5,7 @@ class _LRScheduler:
     def __init__(self, optimizer: Optimizer, last_epoch: int=...) -> None: ...
     def state_dict(self) -> dict: ...
     def load_state_dict(self, state_dict: dict) -> None: ...
+    def get_last_lr(self) -> List[float]: ...
     def get_lr(self) -> float: ...
     def step(self, epoch: Optional[int]=...) -> None: ...
 
@@ -36,4 +37,3 @@ class CyclicLR(_LRScheduler):
 
 class CosineAnnealingWarmRestarts(_LRScheduler):
     def __init__(self, optimizer: Optimizer, T_0: int=..., T_mult: int=..., eta_min: int=..., last_epoch: int=...) -> None: ...
-    def step(self, epoch: Optional[int] = ...) -> None: ...