`Series/Index.nunique` has `dropna` kw to exclude `NaN`. Add the same kw to `.unique` also. As #13979 changes `.unique` to always return `Index`, we can replace `Index._get_unique_index(dropna)` if this option is added.
