Conda package for mgatk and improving error messages

[TomSmithCGAT]

Would you consider making a conda package for mgatk? In the process of trying to resolve #70, I've been going through the code, and have a couple of observations.

1. A lot of the common issues relate to the working environments not being set up appropriates (mainly versions of Java, R package).
2. mgatk often fails silently, e.g no message on the command line. Even the snakemake log file is often not particularly helpful. I've had to resort to print statements to the snakemake run block to work out where the error was occuring.

(1) could be solved by making mgatk available via conda. I expect this would massively reduce the number of issues relating to working environments not being set up appropriately. This would also allow you to remove picard from the bin directory and just include it as a dependency, which is more sustainable.

(2) At the least, it would be helpful to ensure an error is propogated to the command line when the snakemake pipeline fails. I'm not proficient enough in snakemake to say how the errors in the log could be made more informative, but that would be helpful if possible.

[caleblareau]

Thanks for the interest in this-- it seems like there is a community sourced version now. If I can add anything to the official repo just lmk.

I agree the error handling could be better and am working on that now...
https://bioconda.github.io/recipes/mgatk/README.html