wip early working POC background simulation

[Vrixyz]

• early work in progress for bevy task to run physics dimforge/bevy_rapier#596

Questions

This branch proves it's possible to make such simulation in "full" parallel, but currently to simulated time is not correlated to the "real" time, to sync those, it brings up a few important questions:

• outline a strategy for syncing with frames
  • what happens when physics time takes more time than it's allocated time? Should we block ?
    • if blocking, we should be able to retrieve how much we blocked.
    • if not blocking, the "regular" time could be ahead of physics time.
  • what happens when regular update (cpu+rendering) takes more time than physics ?
    • info dumped next (fixed) frame ?
    • wait for the closest frame to expected time to dump info?
• depending on above questions, settle a configuration API
  • a feature flag or runtime setup?