Use per-input `id_field`

[victorlin]

follow-up to nextstrain/augur#1780 (comment)

Replace the pattern of using a global strain_id_field with an id_field per input. Implementation details:

1. Always use augur merge, even for single inputs.

2. Workflow-level merge_inputs.smk should take an id_field per input.

   inputs:
       - name: default
         metadata: <path-or-url>
         id_field: accession
         sequences: <path-or-url>
   
   additional_inputs:
       - name: private
         metadata: <path-or-url>
         id_field: strain
         sequences: <path-or-url>

   augur merge
      --metadata default=metadata1.tsv private=metadata2.tsv
      --metadata-id-columns default=accession private=strain

3. Use augur merge --output-metadata-id-column to customize the output instead of always using the first file's id_field.

   augur merge
      --metadata default=metadata1.tsv private=metadata2.tsv
      --metadata-id-columns default=accession private=strain
      --output-metadata merged.tsv
      --output-metadata-id-column id

Progress

• Simplify configuration of accession as metadata id column zika#110
• phylo: Standard metadata id field pathogen-repo-guide#108
• Support multiple inputs seasonal-flu#331
• Other files to be updated: GitHub query
  • Propagating merge_inputs changes can be made easier with shared module.