Hi Craig,
When you use ImageIntegration this statistical rejection takes place when you create master calibration files or combine light frames. You get to set the degree of rejection by choosing the rejection algorithm and adjusting the sigma rejection thresholds. BPP (since it uses ImageIntegration) permits you the same options. Hotpixels would not be removed by this statistical rejection- because as you say, they are in each bias, dark, and flat. (They would go away for the lights if you dither your data and align on a light signal, thereby shifting the original pixel array).
(Static) Hot pixels then are really only "removed" by measuring their deviation in intensity compared to neighboring pixels- or by a map you create. But this is a separate process (CosmeticCorrection or DefectMap). It seems for your question you thought it was an all-in-one thing?
Just using ImageIntegration (or BPP Calibrate Only with no CosmeticCorrection) and I think you get what you want.
-adam