With this change the genbank parser no longer discards incomplete genes
directly but keeps them as long as they have complete features
annotated.

For example, the PIK3R2 gene is annotated on NC_000019.9 (or a slice) as
4973..>22328 with two RNA entries. One of these, however, is complete so
it would be a shame to discard the entire gene.

This genbank file is incomplete and incorrect anyway, but this was
not the mistake we want to test.

Suppress warning about env variable in tests

Create legend only after gene model enrichment

This fixes a bug where transcripts created from CDS by construction did
not show up in the legend because the legend was created before that
construction.

Fix docstring in sliceChromosome API method

Configuration by environment variable is optional

Parse genbank file without VERSION field

Partial fix for https://humgenprojects.lumc.nl/trac/mutalyzer/ticket/188

Speedup NCBI mapview file import

Instead of querying the existing mappings for overlap and either
updating or inserting depending on the result, we now delete
overlapping mappings first and then only insert.

This is roughly twice as fast. But of course still a horrible
setup compared to some kind of UPSERT functionality which is
unfortunately missing in current PostgreSQL.

Drop BatchJob.download_url column

This is now created on use, by #111.

Fixes #112

Drop TranscriptProteinLink database table

This data is now in Redis, by #94.

Fixes #95

Update requirements

Readable name for batch notification sender

Release 2.0.14

Made e-mail address optional in batch checker website interface.

This address is set with the new `BATCH_NOTIFICATION_EMAIL` config
setting and defaults to the value of `EMAIL`.

Fix counters reporting on about page

We forgot to include the description extractor in the totals. Also,
it said batch jobs were excluded from the totals, which is not the
case.

Log result id for finished batch job

Fix link in batch notification if submitted from webservice