Revamp constructors #57
Description
There are a couple of issues with the phyloseq constructions for otu table, sample data, and tax tables. One has to do with how dummy sample and otu names are created, when the existing names are 1:n; see #53. Another example is that the data.frame tax_table method uses dummy names instead of row names for the taxa names. Given the stalled development of phyloseq, it might be worth completely reimplementing the constructors in speedyseq to have a more flexible and consistent behavior.
- The "sa#" and "ta#" dummy names will not be created unless the relevant (row/col)names is NULL, at least by default. So this would only happen for matrices, not data frames, since the names are never null for data frames. Perhaps have an optional arg that can give the original phyloseq behavior (though I think just giving a warning would be sufficient)
- Have methods that work as expected for data.frame, tibble, and data.table, and perhaps also for Data.Frame, Biom (from the biomformat package), and matrix types from Matrix