Fix/null deletion edge case

[JeremyWesthead]

Give None for gene position in cases where the start of a deletion in a revcomp gene goes past the start of the gene's promoter

[mcolpus]

What would be the impact of this for the katG deletion?

[mcolpus]

Is it easy to add a test case for this just to make sure there aren't any off by one errors, or something like that?

[JeremyWesthead]

It's not particularly easy to add a test case for this - it'd either have to be a new dummy genome or a contrived TB test case (slow). However, it shouldn't be off by one as > should be checking that the calculated nc_num is less than the minimum gene position

[JeremyWesthead]

What would be the impact of this for the katG deletion?

This should result in 0 changes to the prediction, but should change the variants block

        "nucleotide_index": 2156034,
        "gene_name": "katG",
        "gene_position": null,
        "codon_idx": null,

This gives a null value for gene_position rather than a misleading position which doesn't exist