Add zero imputation and use lmer

[lmartinezgili]

Thank you for ANCOM!
I have added a couple of changes which might be of interest:

1. Choose zero imputation instead of adding a pseudocount
2. Use of lmerTest::lmer instead of nlme::lme, for non-nested models

I also suggest changing the function name to lower case

[FrederickHuangLin]

Thanks for your valuable inputs!

How to handle zeros is critical and also controversial in microbiome studies, so I turn to keep that part as it is to make it consistent with the ANCOM paper.

As for switching nlme:lme to lmerTest:lmer, will definitely reflect it in the ANCOM R package (upcoming soon)!

Best,
Huang