biNormal

Compute bivariate normal function

  g = biNormal(xSpread,ySpread,theta,N)

 This does not properly account for a full covariance matrix.  It only has
 different std dev on x and y.  But you can rotate the thing.  The
 ordering is (a) build a bivariate normal aligned with (x,y) and scaled
 by the x and y spreads, (b) rotate the result.

 Example
   g = biNormal(5,10,0,128); imagesc(g), axis image; 
   g = biNormal(5,10,45,128); imagesc(g), axis image; 
   g = biNormal(5,10,45,16); imagesc(g), axis image;