I can certainly understand the temptation. With funding rates so low, odds are that both projects won't be funded at the same time anyway. But this seems like crossing a line to me. If both projects were to be funded, my colleague would either be setting up trainees in direct competition with each other, or else using money earmarked for one thing for something else entirely. It is one thing to use some of a project's budget on interesting side avenues--after all if we knew the outcome already, it wouldn't be research. I've heard that in some fields, it is the norm to have most of the data already (not just preliminary stuff) before submitting proposals, but not in mine. Maybe this is how they do it? It is not like anyone really checks up to make sure that no double dipping is going on.
I do propose strongly related projects that emphasize different aspects of an overall theme, since this is the only sensible way to get enough money to do longterm projects and/or projects that require a lot of resources, especially at my career stage. I also want my students to have clearly delineated and separable projects so there are no issues when it comes to writing up their PhDs. But I try to avoid the temptation to allow the degree of overlap to get too large--if the NSF is paying for something it hardly seems right to charge the DOE for the same thing!
Am I hopelessly naive?