No warnings for collinear effects

[certara-mtomashevskiy]

Dear Helmut!
I think it would be good to add some warnings about matrix rank deficiency to cope the cases when the user's input is not correct

library(replicateBE)
rds01$treatment <- as.factor(sapply(as.character(rds01$period), switch, `1` = "R", `2` = "R", `3` = "T", `4` = "T"))
method.A(data = rds01)
#Error in if (round(res$"CL.lo(%)", 2) >= 100 * ret$BE1 & round(res$"CL.hi(%)",  : 
#missing value where TRUE/FALSE needed`

method.B(data = rds01)
# Error in MEEM(object, conLin, control$niterEM) : 
# Singularity in backsolve at level 0, block 1

method.B(data = rds01, option = 3)
fixed-effect model matrix is rank deficient so dropping 1 column / coefficient
#Error in EMA.B$coefficients["treatmentT", "Estimate"] : 
#  subscript out of bounds
``` 
lm() does not inform about any collinear effects, I think it is better to use 'singular.ok = TRUE' and transform the error to the warning
lme() is also not polite in the error, but we can add more informative error, suggesting to check the data for collinear effects and typos.