Better video timeline segmentation template/support.

**Is your feature request related to a problem? Please describe.**
The current video timeline segmentation template is deficient in a number of very important features, which result in annotations that are hard to control and an overall poor user experience.

**Describe the solution you'd like**
Most of the above features I'm listing are supported by tools such as [VGG VIA](https://www.robots.ox.ac.uk/~vgg/software/via/) or [Vidat](https://github.com/anucvml/vidat).

To name a few, the template could do with the following features:
- support videos without audio
- the ability to automatically detect the FPS of a video without the need to explicitly specify in the template configs
- the ability to jump between temporal segments
- the ability to increase/decrease the extent of the left or right edge of a selected temporal segment for fine-grained annotation
- the ability to play/pause a video while being locked to the segment boundary.
- the ability to reduce playback speed below 0.5 and above 2.5 and control it on the fly with keyboard shortcuts
- the ability to seek either the start or end of a temporal segment while annotating.

**Describe alternatives you've considered**

Video timeline segmentation is a very specific use case. It is akin to annotating all frames in a video that has an event of note, for example, a part of the video where it's raining or snowing, this use case does not translate well to object-detection or tracking templates as alluded [here](https://github.com/heartexlabs/label-studio/issues/1837#issuecomment-993062982) nor does adding a silent track to potentially thousands of videos a viable solution.