Many genbank reference files contain more than one gene, especially
slices from an assembly. Some of these genes may be incomplete in
the reference file (i.e., either start or end exceeds the outer
coordinates). We cannot really do anything with these genes, so we
discard them during parsing.

Given a gene symbol and optional genome build, this returns the location
of the gene.

Primary motivation for this is LOVD, where it will be used in combination
with sliceChromsome as an alternative for sliceChromosomeByGene which only
works on the fixed Ensembl genome build.