Cache `gene_id` mappings w/ reference_data.

[bpblanken]

• Avoid dependency on outgoing URL call.

[tomiles]

I guess this is related to #934 , because we are getting rate limited by the ebi server during large data imports (requests returning Connection refused) so a local cache would help remove this external dependency.

[tomiles]

For now implemented a quick fix by adding a pickle file cache to load_gencode_gene_symbol_to_gene_id() and load_gencode_ensembl_to_refseq_id() that we keep in the reference datasets directory.