Allow user to select multiple data sets for batch execution through UI

[ArtPoon]

For example, executing the mixed HCV pipeline on a MiSeq run's worth of paired FASTQ files.

[ArtPoon]

James proposes implementation of a new widget with tabular display where rows correspond to pipeline inputs and columns correspond to run ordering, or vice versa. Any complicated batch configuration should be done with the restful API. If there are N runs, each input must either have N entries or 1 entry that is replicated across runs.

[ArtPoon]

The plan is to use DataSet filtering to populate columns in a table that correspond to pipeline inputs. User selects groups (or single) DataSets for each pipeline input in the order of those inputs. Table rows are automatically propagated with combinations of DataSets up to the length of the largest group (see above comment).

[ArtPoon]

Subsumed by #488