specifying multiple lineages for `sourmash tax` commands

[bluegenes]

for sourmash tax, specify multiple taxonomy files like so: sourmash tax annotate -g GATHER.csv --taxonomy DB_1.csv DB_2.csv

Here, DB_1 and DB_2 taxonomies will be loaded, with DB_2 overwriting any shared identifiers.

The following command is "valid" but does NOT have the same function:

sourmash tax annotate -g GATHER.csv --taxonomy DB_1.csv --taxonomy DB_2.csv

Here, we only read DB_2 and DB_1 is ignored entirely.

[ctb]

strikes me as a bug. this can be dealt with easily in argparse.

[bluegenes]

Adding action="extend" should handle this nicely. Will PR later today.