Support loading reference files from HDFS/other file system

[fnothaft]

This is supported as part of the ADAM pipe transform using the optional files parameter, which copies files from a "global" path into local temp space on each Spark executor.

[heuermh]

So what needs to be done for this issue? The files parameter in the pipe API is available for every Cannoli cli class to call, do we simply need to add a command line argument flag to add reference file(s) to files? Such already exists in https://github.com/bigdatagenomics/cannoli/blob/master/cli/src/main/scala/org/bdgenomics/cannoli/cli/Bwa.scala#L91

[heuermh]

Fixed by #107.