Nightly Webots simulation setup

[texhnolyze]

Summary

We want to have a working distributed simulation setup for Webots, which allows us to do the following:

• simulate whole games 4 vs 4
• simulate specific setups (e.g. single player playing until first goal is scored)

This should allow us to have more data for training available in general and specific setups, will allow us to have more data which consistently contains active actions (walking, kicking, etc.) instead of standing and might improve bias in our dataset/models.