Use Hadoop FileSystem concat to efficiently merge sharded BAM

[tomwhite]

Hadoop FileSystem supports a concat operation for some implementations (notably HDFS) that can efficiently merge files. (In HDFS it takes advantage of variable block sizes, which avoids the need to rewrite the file.)

SAMFileMerger#mergeParts() could take advantage of this operation if the target filesystem supports it.

[heuermh]

See
https://github.com/bigdatagenomics/adam/blob/master/adam-core/src/main/scala/org/bdgenomics/adam/util/FileMerger.scala

The disableFastConcat flag is necessary because the fast file merger can fail if the underlying file system is encrypted, or any number of undocumented invariants are not met.