fix: order of types

[ShuguangSun]

Try to fix issue #10

Loop through the types as ordered by the useer.

[friendly]

Thanks for this. I'll check it out.

[friendly]

This PR fixes problems 1 & 3 reported by @mattkumar, but it does not fix problem 2. In particular:

## Test 2 - types = "rmeans" [Not OK]
# In this test case, the `row means` test statistic is of particular interest and specify it directly.

CMHtest(Freq~x+y|k, data=r_cmh, overall=TRUE, details=TRUE, types = "rmeans")$ALL$table

# The df and p-value are still returned as NA.

# Chisq    Df Prob
# rmeans 26.02779 NA NA  

I've merged it anyway for now. This case should be investigated further