[FBcode->GH] Cherrypick bkorbar pyavapi (#6963)

* [fbsync] Bkorbar/pyavapi (#6943)

Summary:
* Test: add backend parameter

* VideoReader object now works on backend

* Frame reading now passes

* Keyframe seek now passes

* Pyav backend now supports metadata

* changes in test to reflect GPU decoder change

* Linter?

* Test GPU output

* Addressing Joao's comments

* lint

* lint

* Revert "Test GPU output"

This reverts commit f62e955d7dc81bcb23b40d58ea75413b9b62e76d.

* lint?

* lint

* lint

* Address issues in build?

* hopefully doc fix

* Arrgh

* arrgh

* fix typos

* fix input options

* remove read from memory option in pyav

* skip read from mem test for gpu and pyab be

* fix test

* remove unused import

* Hack to get reading from memory work with pyav

* patch audio test

* gallery change in a hope that docs won't break

* check video decoder inside io

* adding missing lib loading code

* remove unused input

Reviewed By: YosuaMichael

Differential Revision: D41376278

fbshipit-source-id: 1575f63a524227e69187990d88973fc50f705217
Co-authored-by: Bruno Korbar <bkorbar@quansight.com>
Co-authored-by: Joao Gomes <jdsgomes@fb.com>

* cherrypick chages made during fbsync
Co-authored-by: Joao Gomes <jdsgomes@meta.com>
Co-authored-by: Bruno Korbar <bkorbar@quansight.com>

torchvision/io/__init__.py

@@ -4,6 +4,11 @@ import torch
 
 from ..utils import _log_api_usage_once
 
+try:
+    from ._load_gpu_decoder import _HAS_GPU_VIDEO_DECODER
+except ModuleNotFoundError:
+    _HAS_GPU_VIDEO_DECODER = False
+
 from ._video_opt import (
     _HAS_VIDEO_OPT,
     _probe_video_from_file,
@@ -28,7 +33,7 @@ from .image import (
     write_jpeg,
     write_png,
 )
-from .video import _HAS_GPU_VIDEO_DECODER, read_video, read_video_timestamps, write_video
+from .video import read_video, read_video_timestamps, write_video
 from .video_reader import VideoReader
 
 

torchvision/io/_load_gpu_decoder.py

+from ..extension import _load_library
+
+
+try:
+    _load_library("Decoder")
+    _HAS_GPU_VIDEO_DECODER = True
+except (ImportError, OSError):
+    _HAS_GPU_VIDEO_DECODER = False

torchvision/io/video.py

@@ -9,17 +9,9 @@ from typing import Any, Dict, List, Optional, Tuple, Union
 import numpy as np
 import torch
 
-from ..extension import _load_library
-
 from ..utils import _log_api_usage_once
 from . import _video_opt
 
-try:
-    _load_library("Decoder")
-    _HAS_GPU_VIDEO_DECODER = True
-except (ImportError, OSError, ModuleNotFoundError):
-    _HAS_GPU_VIDEO_DECODER = False
-
 try:
     import av