suggested edits to dif_models

[peterhalpin]

Like dif_reports, this function has no required arguments. The function also combines two versions that depend on having different input, which is not user friendly. Let's re-write this function to either version 1 or version 2 below.

version 1:
dif.analysis is provided, the data and the no.dif.mod are passed by dif.analysis, and the user can optionally take the biased.items from dif.analysis.

version 2:
dif.data is provided, in which case no.dif.mod has to be computed (perhaps again) and the biased.items should be a numeric vector indicating the columns of item.data. There should be a helper function that pulls the biased.items from dif.analysis and formats them for input into dif_models, in case the user wants to use the output of dif.analysis for that purpose.

Please also update the documentation accordingly.

[knickodem]

For users interested in supplying a numeric vector to biased.items, there is no reason for running dif_analysis. Instead they can supply the necessary inputs directly via dif.data. There is no case where both dif.analysis and dif.data are needed.

Eventually, we create an dif.data.prep class and a dif.analysis class. Then dif_models only needs 2 arguments - input and biased.items - dif_models pulls the required information based on the class of the input.

[peterhalpin]

I'll write dif_models2 for version 1. Can leave dif_models as is.

[knickodem]

I really don't think there is a need for two functions. dif_models has been changed to only have dif.analysis and biased.items arguments. If people get annoyed that they have to run dif_analysis first, then we can make slight alterations to also allow for dif.analysis argument to also accept a dif.data object.

[peterhalpin]

PFH to write dif_models; dif_models2; update help documentation; Please leave open until I get to this, otherwise I will forget to check it when I come back to this project.