These unit tests should be improved

## test-standard-errors

- Standard error of variance

## test-ci_cor

- ci_cor(X, type = "boot"): There is a comparison with an online example, but I'd
love to see better/more

## test-ci_proportion

- Bootstrap CI consistent with the one for the mean. No test for "stud" though (other formula for standard error used)

## test-ci_measures_of_scale

- ci_IQR/ci_mad are not explicitly tested with other implementation (Bootstrap)

## test-ci_location_shift

- ci_quantile_diff: only tested by keeping one vector constant 3 (Bootstrap)

## test-ci_rsquared()

- There is no test that the estimated NCP of the F distribution is correct, i.e.,
that f_to_ncp() does the right thing. Reference for the formula is Smithson, p 38.

## test-ci_cramersv()

- No explicit check that Bootstrap CIs of ci_cramersv() and ci_chisq_ncp() are correct
