Open closed phases

[masonlr]

This pull request represents work undertaken to support multiple challenge phases.

[masonlr]

The reference to a custom folder "./noise_sigma0.1_modified_workbook_finalized_env.py_DDPG/" shouldn't be in the code: https://github.com/rangl-labs/netzerotc/blob/016432cfe794049c8f54241c1a417b1cd455108a/local_agent_training_and_evaluation/train.py#L13

[masonlr]

Have refactored so there is now a single rangl package with all three environments, you can now use:

gym.make("nztc-dev-v0")
gym.make("nztc-open-loop-v0")
gym.make("nztc-closed-loop-v0")

There is no need to duplicate any of the server logic as the agent can specify which gym environment it wants to use, for example with this code

https://github.com/rangl-labs/netzerotc/blob/f26b9bd0583c715bd20216df0a0f6d63fa19b3a0/random_agent_submission/agent.py#L11-L13

you can state the version of the environment that you want to use:

# env_id = "nztc-dev-v0"
# env_id = "nztc-open-loop-v0"
env_id = "nztc-closed-loop-v0"

[masonlr]

Merging the YAML lists gives:

This way we can submit to multiple challenge phases.

[masonlr]

Can submit to multiple phases and view their respective leaderboards:

[moriartyjm]

The reference to a custom folder "./noise_sigma0.1_modified_workbook_finalized_env.py_DDPG/" shouldn't be in the code:

https://github.com/rangl-labs/netzerotc/blob/016432cfe794049c8f54241c1a417b1cd455108a/local_agent_training_and_evaluation/train.py#L13

100 fixes this

[masonlr]

Closing, as this has been replaced by https://github.com/rangl-labs/netzerotc/pull/101