This has been discussed before.
The primary problem is that the mic input can't sample high frequencies, so it would have to guess at those.
Also, a perfectly flat response isn't necessarily what you'd want. Although it would be an interesting starting point, from which you (or the calibration program) could apply more ear-pleasing curves.
Although I think both of those issues could be overcome with the proper software. My dream would be a program which puts the EQ into 4-way parametric mode, then calibrates test tones through each of the 4 speakers individually.
Question: Does anyone have the programming knowledge required to sample from the mic right now? If someone could get something working which spits out data, then it would be a start in that direction.
If you think about it, once a calibrated mic is chosen, it's purely a software problem...
___________
Tony Fabris