When executing a StepGroup, only load required datasets and scalars
In each version of generateInputCode (there's one for each script language),
only generate code to load the data sets that are actually used by the current
StepGroup. Right now all datasets input to the pipeline are loaded for each
Note that this is done correctly for the Scalar inputs in the Matlab plugin,
but this breaks the MatlabPluginTest of generateInputCode, as there is no
StepGroup setup in the test.
Updated by Rich Williams over 20 years ago
Changed behavior so only data files used at some place in the pipeline are
loaded. This doesn't cover every case of loading unnecessary data, but does
exclude the common case of selecting a dataset that contains two data tables,
but only processing data from one of the tables.