Add variant effect ontology

[fnothaft]

Suggested by @heuermh, see here for examples. This should be added to the VariantAnnotations and variant annotation parser.

[heuermh]

My further suggestion would be to only accept SO term names in ADAM and not worry about mapping between non-SO terms as described on that GEMINI documentation page, i.e. only those values from the column Sequence Ontology terms (column: impact_so) in this table

http://gemini.readthedocs.org/en/latest/content/functional_annotation.html#so-impact-definitions-in-gemini

Although on review that column has some errors in it, e.g. feature elongation should be feature_elongation (SO:0001907), non_coding_exon_variant should be non_coding_transcript_exon_variant (SO:0001792), nc_transcript_variant isn't a valid term, etc.

[heuermh]

I have been able to fix some of those documentation errors, not sure yet if the GEMINI code also needs to be fixed.

arq5x/gemini#300

In any case, the SO itself should be the source of record for these terms.

[heuermh]

See new VCF annotation standard 'ANN' field specification, from authors of SnpEff and Ensembl VEP.

[fnothaft]

See new schema in bigdatagenomics/bdg-formats#67.

[fnothaft]

Closing as dupe of #1044.