utils.load_features cannot parse Nextclade gene map GFFs

[huddlej]

Current Behavior

Nextclade provides gene map GFFs for its many datasets. These gene maps use the gene_name qualifier key to define the name of each gene (e.g., the SARS-CoV-2 gene map). Augur's utils.load_features function currently only checks for gene names stored with qualifier keys of gene and locus_tag. When it tries to parse a Nextclade GFF, the load_features function fails to find the gene qualifier, defaults to locus_tag, and then crashes with a key error when that qualifier does not exist.

Expected behavior

We should be able to load features from GFF3 files with gene names referenced by alternate qualifier keys.

How to reproduce

Steps to reproduce the current behavior:

1. Download the latest SARS-CoV-2 gene map.

curl -O https://raw.githubusercontent.com/nextstrain/nextclade_data/master/data/datasets/sars-cov-2/references/MN908947/versions/2022-07-12T12%3A00%3A00Z/files/genemap.gff

2. Import the load_features function from a Python terminal and call it with the gene map GFF.

>>> from augur.utils import load_features
>>> load_features("genemap.gff")
Traceback (most recent call last):
  Input In [4] in <cell line: 1>
    load_features("genemap.gff")
  File ~/miniconda3/envs/nextstrain/lib/python3.9/site-packages/augur/utils.py:166 in load_features
    fname = feat.qualifiers["locus_tag"][0]
KeyError: 'locus_tag'

Possible solution

1. Add a unit test for utils.load_features that attempts to load features from a Nextclade GFF.
2. Add logic to the load_features function to check for gene_name among the other default qualifiers.

Additional context

Related to #953. Although we plan to reimplement load_features with a more modern GFF parser, the solution proposed here will fix this minor but blocking bug in the short term.

[huddlej]

This was resolved by #1017.