A test for same results with internal estimation of error

[dimalvovs]

It is assumed that the error is estmated as argmin(0.1, data*0.1).

Let's implement a test that checks that the both versions are equal.