Replace cat with seqkit rmdup

This better reflects the duplicate handling used by augur merge. In case
of duplicates, augur merge will keep the last one from the input list.
seqkit rmdup keeps the first one from the input list so the order is
reversed.

There are some behavior changes, none of which should have any impact on
downstream usage:

1. The order of sequences in the file is reversed on a file level.
2. seqkit's default behavior wraps lines at 60 characters.

phylogenetic/rules/merge_sequences_usvi.smk

@@ -57,7 +57,7 @@ rule append_usvi:
         metadata = "data/metadata_all.tsv"
     shell:
         """
-        cat {input.sequences} {input.usvi_sequences} > {output.sequences}
+        seqkit rmdup {input.usvi_sequences} {input.sequences} > {output.sequences}
 
         augur merge \
           --metadata ingest={input.metadata} usvi={input.usvi_metadata} \