Modulify and big refactor

[caskok]

• Extended core to include the abstract class BurnInStrategy. This is then implemented by FaceDetectionStrat, TextDetectionStrat and ShotDetectionStrat. Future new type of burn-in implementations are easier this way and this makes the code more DRY.
• Extended core with the function burn_in_video(strat: BurnInStrategy, ... ). This is the general core functionality of the three burn-in detection types: face, text and shot. Specific details are hidden away in their own implementations.
• Added all the types of detections to the module (face, text, shot, midroll, still picker) each with their own interface (similar to the one you describe). The timeline is added as a flag in this interface for shot, text and face.
• Refactored the timeline file to a class which creates the timeline animation and merges it with the burn-in video.
• Added some more documentation in the form of docstrings, type annotations and comments.

[caskok]

@gabriben Do you want to take a look at this or can I merge?