Dropping parameters significant_threshold and fill_missing_values

[stemangiola]

In the process of adding the prefix argument to test_differential_abundance for allowing easy multi method testing I was thinking that we could trim down some of the argument to make the interface more clean

• significant_threshold and significant column. It does not add much as it is pretty easy to filter based on FDR, and if anything make the users think less about what they are filtering
• fill_missing_values we have already fill_missing and impute_missing routines, I was thinking throu a error/warning if needed and let the user operate externally to this function

[mblue9]

That makes sense to me. It had crossed my mind before that the significant column wasn't essential and that it might be better to let them explicitly choose the FDR to filter on to make the threshold clearer.
Yes fill_missing_values could go too I think.

[stemangiola]

We have to keep in mind that this change (although I will use the deprecate framework) will break previous code that use the column significant.

But better taking this decision at this stage than when tidybulk will be more established.