Fix: Multiple header lines in the genbank tsv file

The fetch of rsv genbank data involved three distinct curl calls,
leading to the creation of multiple header lines in the fetched data.
Although the first header line was correctly interpreted, the 2nd and
3rd header lines transformed into inaccurate metadata and sequence
records within the data/genbank.ndjson file. This caused failures during
the subsequent data processing in the transform rule.

This commit addresses the above issue with tsv-append to concatenate the
three files, ensuring that only a single header line is used.

ingest/workflow/snakemake_rules/fetch_sequences.smk

@@ -23,11 +23,14 @@ rule fetch_from_genbank:
     shell:
         """
         curl "{params.URL_a}" --fail --silent --show-error --http1.1 \
-             --header 'User-Agent: https://github.com/nextstrain/rsv ([email protected])' >> {output}
+             --header 'User-Agent: https://github.com/nextstrain/rsv ([email protected])' >> {output}_a
         curl "{params.URL_b}" --fail --silent --show-error --http1.1 \
-             --header 'User-Agent: https://github.com/nextstrain/rsv ([email protected])' >> {output}
+             --header 'User-Agent: https://github.com/nextstrain/rsv ([email protected])' >> {output}_b
         curl "{params.URL_general}" --fail --silent --show-error --http1.1 \
-             --header 'User-Agent: https://github.com/nextstrain/rsv ([email protected])' >> {output}
+             --header 'User-Agent: https://github.com/nextstrain/rsv ([email protected])' >> {output}_general
+
+        tsv-append -H {output}_a {output}_b {output}_general > {output}
+        rm {output}_a {output}_b {output}_general
         """
 
 rule csv_to_ndjson: