Indeed, you cannot run DataTools from a pipeline. The pipeline model of CellProfiler is constructed such that each image is processed independently (with a few exceptions, like CalculateStatistics). You could use the DisplayHistogram module, for example, to display a histogram from an image on each cycle, but to do any pre- or post-processing on all the images, you need to run a separate pipeline or DataTool. I agree with your implication that automating this sort of post-processing would be desirable, and We That Are CP have discussed new models like this for the next incarnation of CellProfiler, though it is still in development so I can’t say whether this will occur or not.
However, anyone is free to write their own “SuperHistogram” module which takes all cycles’ measurements into account and post it here!