Degree of freedom for variance calculation

I found that np.var function is used to calculate the variances in the code. Since the np.var function is using `ddof=0` by default, we need to fix it to use `ddof=1` to make all the equations aligned with the original paper.

I created a PR for it. Please take a look and let me know if you have any thoughts. 

PR link: https://github.com/epigenelabs/pyComBat/pull/30