Update on "[Doc] Add deprecated autocast comments for doc"

# Motivation
We generalize a device-agnostic API `torch.amp.autocast` in [#125103](#125103).  After that,
- `torch.cpu.amp.autocast(args...)` is completely equivalent to `torch.amp.autocast('cpu', args...)`, and
- `torch.cuda.amp.autocast(args...)` is completely equivalent to `torch.amp.autocast('cuda', args...)`

no matter in eager mode or JIT mode.
Base on this point, we would like to deprecate `torch.cpu.amp.autocast` and `torch.cuda.amp.autocast` to **strongly recommend** developer to use `torch.amp.autocast` that is a device-agnostic API.

cc jgong5 mingfeima XiaobingSuper sanchitintel ashokei jingxu10

[ghstack-poisoned]

torch/cpu/amp/autocast_mode.py

@@ -1,4 +1,5 @@
 from typing import Any
+import warnings
 
 import torch
 
@@ -22,6 +23,7 @@ def __init__(
             self.device = "cpu"
             self.fast_dtype = dtype
             return
+        warnings.warn("torch.cpu.amp.autocast(args...) is deprecated. Please use torch.amp.autocast('cpu', args...) instead.")
         super().__init__(
             "cpu", enabled=enabled, dtype=dtype, cache_enabled=cache_enabled
         )

torch/cuda/amp/autocast_mode.py

@@ -1,5 +1,6 @@
 import collections
 import functools
+import warnings
 
 import torch
 
@@ -31,6 +32,7 @@ def __init__(
             self.device = "cuda"
             self.fast_dtype = dtype
             return
+        warnings.warn("torch.cuda.amp.autocast(args...) is deprecated. Please use torch.amp.autocast('cpu', args...) instead.")
         super().__init__(
             "cuda", enabled=enabled, dtype=dtype, cache_enabled=cache_enabled
         )