Use rxode2::etTrans() instead of nlmixr_data_simplify()

[billdenney]

@mattfidler, In an initial test, rxode2::etTrans() drops the MDV column. This could be important for nlmixr2targets in case the MDV column changes. Mainly, I want to be sure that all columns are captured. Since etTrans() is an rxode2 function, might it miss any columns that are estimation-focused rather than integration-focused.

[mattfidler]

@billdenney You are right. MDV is not used by nlmixr2 or rxode2, it simply translates it to an EVID equivalent.

As I was thinking, there are a few things that may change though:

• IDs may change, they are saved as factor information in etTrans()
• Time varying covariates and non-time-varying covariates are separated (and handled differently).

Perhaps it is easier to simply use nlmixr_data_simplify()

[mattfidler]

(Mostly for the ID issue. Also using etTrans directly has not been widely tested so there may be some edge cases that I am not anticipating).

[billdenney]

It sounds like long-term solution could be to have this in rxode2::etTrans(), but the today solution is to use nlmixr_data_simplify().