Interesting you should post this!
I've been considering such a test for the past month and I'm now trying to attain all the necessary tools for such..
The idea behind this has been to test the statement, Exposure controls density, but development controls density and contrast. I'm not try to test the veracity of this, but the quality and extent of manipulation possible.
Yes I am very well aware about the variations in emulsion and developer combos but in reality I'm not interested in discovering a universal theory of everything photographic. Just the developer/film combo I'm interested in.
Films I like PanF+ and HP5/Tri-X (decided eventually to stick with HP5). Developer available to me that works nicely with both? Rodinal (Adonal to be more specific).

Let the games Begin..
The general parameters behind this is set, I will construct a standard and controlled scene (which will include a target). The target area's lighting will be metered as zero at full box speed, then extensive bracketing will take place.
The idea is.. Photograph a control scene at several different exposure indexes. Say for example, -4,-3,-2,-1,0,+1,+2.
Repeat this several times and develop each strip for different a different amount of time. Say for example. -60%,-30%,-15%,0 (manufacturers suggestion), +15%,+30%,+60%.

I am quite aware that there is some waste involved with this. As a scene that is seriously under exposed and under develop will produce little to no results. But the idea is to produce a scientific test, or as scientific as possible. So temperature, agitation, general processing routine, scanning (of course), and all other variables be standardized.

Hopefully once I'm done, I'll have a menu of options indicating the exact look I want for a given developer/film combo.
What I do need help with is, how much of each.
I decided quite easily that I should vary the exposure in one stop increments, for simple ease of application afterwards. I wanted to maintain waste to a minimal and my idea would be to have 6 exposures per development strip. HP5 rated at 3200 (-3), 1600 (-2), 800 (-1), 400 (0), 200 (+1), 100 (+2), for example. PanF+ rated at 400 (-3), 200 (-2), 100 (-1), 50 (0), 25 (+1), 12 (+2). (even though PanF+ rated at 400 sounds really silly)
But, how much should I vary ("bracket") development time?? Standard for 0 (iso 400 and 50) and how much plus or less? I've heard figures like 15% for each stop but sometimes that doesn't look right.. Maybe it's not a linear function? I would rather it be a constant between both films but I'm open for whatever works!

Any help will be VERY welcome! Thoughts, comments or criticisms as well!