test function is insensitive for NaN vs NA difference #2960

For differentiating NA from NaN currently behavior of test function is consistent to all.equal while more appropriate IMO would be to expect identical behavior.

all.equal(NaN, NA_real_)
#[1] TRUE
identical(NaN, NA_real_)
#[1] FALSE

Unfortunately it is not that simple, we would need to overwrite all.equal.numeric to retain current features that comes from all.equal and be able to differentiate NA from NaN.

This ideally should be fixed upstream in R giving an option to differentiate both.