Clarify train sim v3 documentation

[Castavo]

Already work in progress in #168

Things to improve:

• add non goals
• better goals
• prioritize goals
• translate diagrams from the slideshow to english, add these here
• explain how driver behavior modules work exactly
• give examples of driving instruction sequences for each kind of constraint
• describe what APIs driver behavior modules need to simulate samples
• give high level APIs for driver behavior modules and driving instructions
• give examples of corner cases for driver behavior modules, especially for the "reconciliation step"
• describe the exact external FFI api, and its differences from the rust library API
• add a short description of concepts, actors and who does what
• list parameters present in a train state
• challenge the assumptions which would allow "combining constraining parameters" (step of the driver behaviour module process)
• enrich design specifications in _index.en.md
• the distinction between abstract and concrete driving instructions could be removed
• Think a bit harder about how the "slowdown coefficient" would be updated (describe how driver behavior modules would use the slowdown coefficient to monotonically generate decisions)
• Also think a bit harder about the loop over steps

Acceptance criteria

• Design decisions are properly justified
• Examples are given on corner cases
• Overall understandable by a newcomer (maybe upgrade objective to "a newcomer could implement it" ?)