digits hilbert_inverse.adb accuracy test
 50-digit floating point, Hilbert vs random matrix
for N by N matrices, N=2, 4, 8, 16, 32, 64, 128, 256
(this may hit storage limitations)

N= 2, norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 1.00000000000000E-50, maxerr= 2.00000000000000E-50
random  norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 5.00000000000000E-51, maxerr= 1.00000000000000E-50

N= 4, norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 5.93750000000000E-48, maxerr= 8.50000000000000E-47
random  norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 5.73125000000000E-51, maxerr= 1.00000000000000E-50

N= 8, norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 4.19720156250000E-42, maxerr= 5.65263000000000E-41
random  norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 7.24781250000000E-51, maxerr= 2.23570000000000E-49

N= 16, norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 3.39435846280965E-30, maxerr= 4.33854232016800E-29
random  norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 8.94609374999999E-50, maxerr= 7.26000000000000E-48

N= 32, norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 1.66738856722820E-06, maxerr= 4.91046020088051E-05
random  norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 1.17366337890625E-50, maxerr= 1.60300000000000E-49

N= 64, norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 3.48089330774612E+00, maxerr= 4.74511972590516E+01
random  norm= 0.00000000000000E+00, mindiag= 1.00000000000000E+20
     avgerr= 6.03160229980470E-50, maxerr= 5.61000000000000E-49