In-part / non-unique names

[tpoisot]

This closes #39 by erroring when there are multiple nodes with the same name, as in in-part names.

[tpoisot]

New issue: there are some cases where a genus and subgenus share a name, like Mus.

julia> NCBITaxonomy.names_table[findall(isequal(10088), NCBITaxonomy.names_table.tax_id),:]
3×4 DataFrame
 Row │ tax_id  name    unique_name  class
     │ Int64   String  String?      NCBIName…
─────┼────────────────────────────────────────────────────
   1 │  10088  mice    mice <Mus>   class_common_name
   2 │  10088  mouse   mouse <Mus>  class_common_name
   3 │  10088  Mus     Mus <genus>  class_scientific_name

julia> NCBITaxonomy.names_table[findall(isequal(862507), NCBITaxonomy.names_table.tax_id),:]
1×4 DataFrame
 Row │ tax_id  name    unique_name     class
     │ Int64   String  String?         NCBIName…
─────┼───────────────────────────────────────────────────────
   1 │ 862507  Mus     Mus <subgenus>  class_scientific_name

This might require to add something like the rank in GBIF

[tpoisot]

Alternatively, we can drop a keyword argument to taxon which would be an instance of NCBINameClass (or whatever it's called).

[tpoisot]

The best solution is probably to stick to the GBIF.jl API and pass the rank when there is an ambiguity. So this would be an additional error type NameRankAmbiguous), with a list of possible levels.