[fvsc] in ggttg mad, set JAMP2(0) in cudacpp-only mode where SMATRIX1…

… is not called (dummy random choice of color madgraph5#402)

This is a WORKAROUND for "Error: failed to reduce to color indices" oliviermattelaer/mg5amc_test#14

Note: in this patch I only set JAMP2(0), while all other JAMP2(1:NCOLOR) are zero.
It seems that addmothers.f already has a workaround to cover this case.

epochX/cudacpp/gg_ttg.mad/SubProcesses/P1_gg_ttxg/auto_dsig1.f

@@ -564,6 +564,12 @@ SUBROUTINE SMATRIX1_MULTI(P_MULTI, HEL_RAND,  CHANNEL, OUT ,
           OUT(IVEC) = OUT2(IVEC) ! use the cudacpp ME instead of the fortran ME!
         END DO
       ENDIF
+
+      IF( FBRIDGE_MODE .EQ. 1 ) THEN ! (CppOnly=1 : SMATRIX1 is not called at all, JAMP2_MULTI is not filled)
+        DO IVEC=1, NB_PAGE_LOOP
+          JAMP2_MULTI(0,IVEC) = NCOLOR ! workaround for https://github.com/oliviermattelaer/mg5amc_test/issues/14
+        END DO
+      ENDIF
 #endif
 
       IF ( FIRST_CHID ) THEN