partly revert last commit: only real input vector get special scaling for dc component

src/maths/cmaths/cmath4.c

@@ -713,9 +713,7 @@ cx_fft(void *data, short int type, int length, int *newlength, short int *newtyp
         outdata = alloc_c(fpts);
 
         scale = (double) fpts;
-        outdata[0].cx_real = out[0][0]/scale/2.0;
-        outdata[0].cx_imag = out[0][1]/scale/2.0;
-        for (i = 1; i < fpts; i++) {
+        for (i = 0; i < fpts; i++) {
             outdata[i].cx_real = out[i][0]/scale;
             outdata[i].cx_imag = out[i][1]/scale;
         }
@@ -746,9 +744,7 @@ cx_fft(void *data, short int type, int length, int *newlength, short int *newtyp
         outdata = alloc_c(N);
 
         scale = (double) N;
-        outdata[0].cx_real = datax[0]/scale/2.0;
-        outdata[0].cx_imag = datax[1]/scale/2.0;
-        for (i = 1; i < N; i++) {
+        for (i = 0; i < N; i++) {
             outdata[i].cx_real = datax[2*i]/scale;
             outdata[i].cx_imag = datax[2*i+1]/scale;
         }