Consolidate approaches to render WebVTT subtitles

[dsilhavy]

Description

We are currently supporting two different ways to render WebVTT subtitles. The default approach relies on native events of the platform such as cue.onenter and cue.onexit . Moreover, the rendering of the subtitles is handled by platform itself.
As a fallback solution for platforms that do not support native events and rendering, we added vtt.js and dispatch the required timing events ourselves. Both approaches also use their own WebVTT parser.

We should aim to consolidate this and move towards a complete custom rendering solution such as the one we implemented with vtt.js. First tests also indicated that the parser we use for custom rendering is faster than the one we use for native rendering.