How hard/expensive would it be to build/buy a device to measure the force of attaching and detaching plates and bricks?
Thankfully we don't need to think too hard on that, as pcantin has already built a similar device, admittedly built for measuring the repetitions it takes for a brick to lose its clutch power. But it should not be too hard or expensive to modify that design to include something that can measure the exact forces involved. Or it maybe possible to measure the amount of current required by the servo during attaching or detaching and to infer the involved forces from that and thus solve the problem witout requiring any major hardware alterations.
But if your goal is to compare a number of various bricks, I'd recommend a different route. Attach a number of nominally same bricks (i.e. a bunch of 2x4 bricks for example) together in a stack, then pull the two end bricks apart. This way you are applying some force to the top and bottom bricks which gets transmitted across all intermittent bricks. By increasing the applied force the two bricks with the weakest bind will separate first, be they anywhere in the stack. This is akin to finding the weakest link in a chain by pulling the end links apart to see which link break first.
Of course this will only tell us which connection (consisting of a brick's stud-side and another brick's antistud-side) has the weakest clutch power, but by repeating the same test with varied composition (all bricks from manufacturer A, the same bricks in another order, bricks from A mixed with bricks from B, bricks from A on the top and from B on the bottom, etc.) with some planning and measuring it should be possible to establish intra-manufacturer and inter-manufacturer variations in binding force.