why do you think a 32-bit representation will be necessary/helpful?
I've been told that tone-mapping before stitching can cause artifacting and stitching errors, depending on the tonal makeup of the scene. Because my display, and I suspect everyone else's display can show 32-bit depth images, I don't see any good reason to tonemap the HDR files (which will compress them to 16-bit files) BEFORE completing all the editing I want to do with them(i.e. focus stacking and panorama stitching).
The problem is that for HDR+Focus stacking+panorma projects (I get the feeling no one on earth does this), focus stacking software, such as Helicon, do no take 32-bit files. This throws a wrench in my planned workflow. I haven't actually tried this stuff, by the way, but am researching for the optimal workflow and tools to use. It's a bit complicated to say the least.