[Bug] mmcv.VideoReader needs to provide an async API to handle large videos

Prerequisite

[X] I have searched Issues and Discussions but cannot get the expected help.
[X] The bug has not been fixed in the latest version(https://github.com/open-mmlab/mmcv).

Environment

When the video is large - it makes no sense to read the whole video into memory like video = mmcv.VideoReader(video_file) does. It quickly runs out of memory.

mmcv.VideoReader should provide an async API instead.

In C++ this would look something like this:

mmcv::VideoReader(video_file, [](const VideoFrame &frame) {
    // process the frame
});

Reproduces the problem - code sample

n/a

Reproduces the problem - command or script

n/a

Reproduces the problem - error message

n/a

Additional information

n/a

open-mmlab / mmcv

[Bug] mmcv.VideoReader needs to provide an async API to handle large videos #3010