Use only the 21L sequence from references

data/references_sequences.fasta contains both `Wuhan/Hu-1/2019` and
`21L`. The aligned input is expected to contain `Wuhan/Hu-1/2019` so the
merging with references_sequences.fasta is really only for 21L.

nextstrain_profiles/nextstrain-gisaid-21L/prefilter.smk

@@ -92,7 +92,8 @@ rule gisaid_21L_aligned:
         exec 2> {log:q}
 
         < {input.references:q} \
-          zstd \
+          seqkit grep --by-name --pattern 21L \
+        | zstd \
         > {output.aligned}
 
         < {input.aligned:q} \