Allow user to select a column from a 2D dataset

[chuckwondo]

Currently, a user may select only 1D datasets (e.g., quality_flag, sensitivity, etc.), but there are cases where a user may wish to select a column (or multiple columns) from a 2D dataset within a granule file (.h5).

For example, a user might wish to select rh100, but the file might contain a 2D dataset named rh with many columns, so specifying rh100 should select rh[100] (i.e., the column at index 100 [0-based index, thus 101st column]) as a 1D "dataset" from the 2D dataset named rh.

Note that the granule files in some collections may already contain rh100 (for example) as a 1D dataset, while files in other collections may not. Unfortunately, this is inconsistent.

However, a user should have a consistent experience without having to know whether or not the column the user wants to select is a 1D dataset or a column within a 2D dataset. In either case, the user should be able to specify, for example, rh100 and get either the 1D dataset that is already within the file, or the column at index 100 from the 2D rh dataset, if no such 1D dataset already exists. In either case, the user simply specifies rh100.

Further, the user may specify as many such datasets as desired (e.g., rh20, rh50, rh100) and the algorithm will do the "right thing" to get the data (either pull existing 1D datasets by the specified names, or pull the corresponding columns from the corresponding 2D dataset).

More specifically, when a user specifies a dataset with a name of the form <NAME><INTEGER>, the algorithm will first attempt to select a 1D dataset with the name <NAME><INTEGER>. If not found, the algorithm will look for a 2D dataset with the name <NAME>, and extract the column at index (0-based) <INTEGER>.

Acceptance Criteria

• when a user specifies a column (dataset) name of the form <NAME><INTEGER>, the algorithm will first attempt to select a 1D dataset with the name <NAME><INTEGER>. If not found, the algorithm will look for a 2D dataset with the name <NAME>, and extract the column at index (0-based) <INTEGER>

[chuckwondo]

Addressed by #16