Introduce performance evaluation

[BartekCupial]

Overview

This PR introduces a new script, eval.py, designed for faster evaluation using multiple environments and leveraging the efficiency of the Sample Factory sampler. Additionally, an example usage of eval.py is demonstrated in the added eval_mujoco.py script within sf_examples.

Key Changes

• New Script: eval.py

  • A fast evaluation script similar to enjoy.py.
  • Utilizes multiple environments for improved speed.

• Example Usage: eval_mujoco.py

  • Demonstrates how to use the new eval.py script.

• Core Logic in evaluation_sampling_api.py

  • Heavily inspired by simplified_sampling_api.py.
  • Main differences is usage of msg_handlers, episode message processing, and loading pretrained checkpoints.

• Additional features

  • Because during the evaluation we might want to use different arguments then during training I've added checkpoint_override_defaults. Has the same purpose as load_from_checkpoint(cfg), but keeps the option of overriding cfg with argv.
  • Added EpisodeCounterWrapper, which counts the episodes for each environment. Used during eval in order not to bias the results with shorter episodes.
  • Save the eval results to a csv file

[BartekCupial]

I'm getting this message when the script finishes, not sure yet why.

[W CudaIPCTypes.cpp:15] Producer process has been terminated before all shared CUDA tensors released. See Note [Sharing CUDA tensors]

[BartekCupial]

@alex-petrenko can I ask for the review, please?

[codecov-commenter]

Codecov Report

Attention: 318 lines in your changes are missing coverage. Please review.

Comparison is base (6379cf9) 79.52% compared to head (a5fde39) 76.32%.

Files	Patch %	Lines
...e_factory/algo/sampling/evaluation_sampling_api.py	0.00%	194 Missing :warning:
sample_factory/eval.py	0.00%	84 Missing :warning:
sf_examples/mujoco/fast_eval_mujoco.py	0.00%	16 Missing :warning:
sample_factory/cfg/arguments.py	7.14%	13 Missing :warning:
sample_factory/envs/env_wrappers.py	28.57%	10 Missing :warning:
sample_factory/envs/create_env.py	66.66%	1 Missing :warning:

:exclamation: Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

```diff @@ Coverage Diff @@ ## master #288 +/- ## ========================================== - Coverage 79.52% 76.32% -3.20% ========================================== Files 97 100 +3 Lines 7517 7845 +328 ========================================== + Hits 5978 5988 +10 - Misses 1539 1857 +318 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[klyuchnikova-ana]

I'm getting this message when the script finishes, not sure yet why.

[W CudaIPCTypes.cpp:15] Producer process has been terminated before all shared CUDA tensors released. See Note [Sharing CUDA tensors]

Yes, this started to happen after some PyTorch update. I've been meaning to investigate, but I also found that it's totally harmless and in practice can be ignored.