Implement observation and reward scaling wrappers

[arnupretorius]

Best practice advice:

• Make sure everything is reasonably scaled.

Rule of thumb:

• Observations: Make everything mean 0, standard deviation 1.
• Reward: If you control it, then scale it to a reasonable value.
• Do it across ALL your data so far.
• Look at all observations and rewards and make sure there aren't crazy outliers

[arnupretorius]

For inspiration:

• https://github.com/openai/gym/tree/master/gym/wrappers
• https://github.com/deepmind/acme/tree/master/acme/wrappers
• https://github.com/openai/gym/blob/a5a6ae6bc0a5cfc0ff1ce9be723d59593c165022/gym/core.py#L262

[KaleabTessera]

Currently investigating if we can just use PZ observation and reward scaling wrappers.