VERSION UPDATE: v2.1

VERSION UPDATE: v2.1

Augmentations:
* Stable conda environment, latest version of py and tf
* Better hash functions for reproducibility
* Params and data now in custom classes
* Support in model for multiple rnn layers, multiple dense layers

REPRODUCIBILITY RESET:
	- New conda environment changes reproducibility. I confirmed this several times running side by side clones of the repo with only env different
	- Original repro case was RTMA data, new case HRRR
	- Original repro used improper scaling using data from prediction phase.
	- New Repro case uses minmax scaling only calculated from training data
	- New hash functions, old ones were hacky
	- New Repro case with more predictors to make sure reproducibility is tracking all potential changes
	- New Repro case with more params engaged: validation data, dropout, etc.
	- New Repro case that had interpolation performed on it
	- New Repro case multi layers

CLEANUP
    - Removing old reproducibility files and support for original repro. To see old code and reproduciblity results, see commit "911c6d7a" or PR#11.