Add unit tests to ensure valid example configs

[jon-tow]

This PR adds the following configuration and example updates:

• Adds unit tests to ensure future config changes do not break examples. Note that this requires TRL config YAMLs to be located in directories named configs. (This might need to be updated if we decide to integrate OmegaConf/Hydra for config management.)
• Adds method config options, scale_reward, ref_mean, ref_std, and cliprange_reward to appropriate PPO configs.
• Fixes examples/experiments/grounded_program_synthesis missing dataset directory error.

Awaiting #109

[LouisCastricato]

This looks great! Does it work with GitHub CI?

[jon-tow]

Yup! I just double checked CI (see here)