MOQ learning

[ffelten]

• [x] Refactor MOAlgorithm into MOPolicy and MOAgent to differentiate between multi net and single net (e.g. CN).

Paper: https://ieeexplore.ieee.org/document/6615007 Implementation based on the code sent by Lucas (thank you!).

• [x] Single policy MOQLearning (known weights)
• [x] Multi-policy MOQLearning (outer loop)

I hope the API makes more sense now. Tell me what you think. I relaunched all the existing examples to see if the refactor broke things down. It does look okay!