Remove scenes in favor of tracks and clips

[leodiegues]

Following industry conventions instead of creating new abstractions would be way more beneficial to the project. Standardizing timeline event features would remove much of type-handling code and simplify the interaction with Mosaico. Tracks and clips are old friends of NLE softwares and we should not reinvent the wheel.

At first, we should not see significant changes on rendering time because of these changes, but they should definitely bring more clarity to the codebase.

Still, we had a good point internally when we proposed the existence of scenes: grouping asset references together seemed (and still seems) a good idea and we should make possible to group clips in the same track via some metadata.