[Feature] Add Support for additional SB3 and RLLIB algorithms

[amd-alexcann]

Summary

Schola currently supports PPO, SAC, APPO and Impala for ray and SAC, and PPO for SB3.

Both libraries include additional algorithms that could be supported but currently aren't.

RLLIB

• DQN

SB3

• DQN
• TD3
• A2C
• DDPG

Motivation (optional)

No response

Proposed change

For each algorithm the following should be added to Schola:

• Settings Dataclasses for each algorithm, so they can be configured from the CLI
• [If Necessary] Specific export code for exporting the model architecture to onnx
• [Optional] Unreal Side code to add support when training from the editor

The following should be updated:

• Sb3/rllib train commands to support the additional algorithms

Describe alternatives you've considered

No response

Additional context

No response