Rendering and Video recording for all environments and algorithms

Rendering and Video recording should work with both MuJoCo and RLBench environments and HAC and SB3 algorithms. The user should also be able to distinguish between rendering at test and rendering at train time.

Rendering	HAC	SB3
MuJoCo	train test	train test
RLBench	train test	train test

Video Recording	HAC	SB3
MuJoCo	train test	train test
RLBench	train test	train test

Rendering and recording is now possible for both RLBench and MuJoCo environments and both HAC and SB3 algorithms. The render_args (default=[['none',1],['none',1]]) in main.yaml specify how and when to render during training (first sublist) and testing (second sublist). 'record' is for video, 'display' for direct visualization, 'none' for not rendering at all. The numbers determine the number of epochs after which we render the training/testing. Example: [['display',1],['record',10]] means that we display every training and record every 10th testing run.

However, some combinations won't work:

• MuJoCo: The training is only displayed in the first epoch if the evaluation should be recorded.
• RLBench: 'display' is either on or off. Either everything (train and test and every epoch) is displayed, or nothing.
• RLBench: You have to decide between 'display' and 'record'. 'record' will overwrite 'display'. If e.g. render_args=[['record',1],['display',1]], the training will be recorded but the evaluation won't be displayed.