matrxi inversion in imagery code

The matrix inversion in the imagery code (both i.rectify and i.block.rectify)
uses Cramer's rule for a 3x3 matrix. This was written long before I knew
anything about the proper way to handle matrix inversion.
Since there were only 3 variables it was "easy" to code. However
one of the methods in NRinC could be used (svdcmp is my preference).
Something of this nature needs to be done for i.rectify in any case
if higher order polynomial transformations are to be used.

Thanks for pointing out the problem. Will record the matrix inversion
problem as a true GRASSBUG.

-----------------------------------------------------------------
Michael Shapiro U.S. Army CERL
email: shapiro@zorro.cecer.army.mil Environmental Division
phone: (217) 352-6511 ext 526 P.O. Box 9005
fax: (217) 373-7222 Champaign, Ill. 61826-9005
-----------------------------------------------------------------