Update existing NCBI functions

[stitam]

Related to issue #26, #80, #82.

Changes in ncbi_get_uid():

• Function has a new argument, na_strings. The defined NA strings will be converted to NA-s and will be removed from the query.

Changes in ncbi_link_uid():

• The function can be called multiple times in a sequence. Each call will take the last column as the query and add a new column. This allows for cleaner workflows using pipe operators.

Changes in ncbi_link():

• The internal dispatch has been simplified. Instead of dispatching to a separate function for each from-to combination, some combinations are now dispatched to a generic link function which seems to work for multiple combinations.

Changes in ncbi_download_genome():

• Changed default arguments: the function will download fasta files instead of genbank files and will no longer mirror FTP structure.
• The function can now be piped with the output of ncbi_link() and ncbi_link_uid(), if the last column contains assemblies.

Changes in ncbi_recover_id():

• The function can now recover nuccore IDs as well.

Other:

• Updated documentation and added new tests.