the problem with any kind of weight and balance takeoff warning system (other than the economic issue cynically (but perhaps accurately) referenced above) is that the accuracy of the available systems would be such that either:
1. The system has a false warning rate so high as to render it more of a liability than an asset; or
2. The system would have warning thresholds set so unconservatively as to render the system all-but useless except in the case of errors so gross that massive system failures would be required to get there in the first place.
It's analagous to the "takeoff acceleration" warning systems people keep proposing; the idea is nice, but there's no real "sweet spot" where you could have a warning system that worked when it was supposed, and didn't go off when not required.
There are already conservativisms in takeoff performance and in weight and cg envelopes which account for normally expected variability, and procedures should be designed to account for these two. If those procedures have broken down to the extent of allowing a seriously unsafe loading to develop, the "takeoff weight warning system" might just end up being one more thing the accident report records as being ignored/bypassed or simply inop....