Clarifying ADAMSaveAnyArgs fields

[ryan-williams]

At the very least, one or both of these should probably be renamed? (code):

  /**
   * If true and saving as a legacy format, we will write shards so that they
   * can be merged into a single file.
   *
   * @see deferMerging
   */
  var asSingleFile: Boolean

  /**
   * If true and asSingleFile is true, we will not merge the shards once we
   * write them, and will leave them for the user to merge later. If false and
   * asSingleFile is true, then we will merge the shards on write. If
   * asSingleFile is false, this is ignored.
   *
   * @see asSingleFile
   */
  var deferMerging: Boolean

It seems like asSingleFile now really means "written such that each shard can be merged, i.e. BAM header is not prepended on each partition, only the first one", and deferMerging means "leave the unmerged, headerless partitions in place"?

There are actually 3 possibilities, right? Would an enum be clearer?

[fnothaft]

There are actually 3 possibilities, right? Would an enum be clearer?

I would agree with that. An Enum would SGTM!

[heuermh]

I'd be ok if ADAMSaveAnyArgs went away and the enum moved to SaveArgs in utils:

https://github.com/bigdatagenomics/utils/blob/master/utils-cli/src/main/scala/org/bdgenomics/utils/cli/ParquetArgs.scala#L30

sortFastqOutput also feels misplaced.

[heuermh]

On further consideration, SaveArgs, LoadFileArgs, SaveFileArgs, and ParquetLoadSaveArgs are all rather worthless, since we always redefine INPUT and OUTPUT meta variables so that we can be more specific about supported file types in the description.

[heuermh]

Note there is also disableFastConcat now