First cut of VariantWorks SDK!

Introducing VariantWorks SDK - python framework to simplify the development of deep learning powered variant and consensus calling from reads.

Highlights of first cut

1. Encoder - Read pileup encoder with support for read, base quality, mapping quality data, and reference and allele proposal encodings.
2. I/O - VCF reader for parsing true and false positive variant samples for training.
3. Data Loader - NeMo wrapped data loader implementations for online and offline data loading/encoding.
4. Samples - A trainer/evaluator app for simple SNP zygosity calling using a CNN.

More details available at https://clara-parabricks.github.io/VariantWorks/