Add missing array initializations to eigen_soc/sorad.f90

The initializations are irrelevant for the production version but circumvent an erroneous
stop due to missing initializations in the debugging version. The commit hopefully
fixes issue #292.

eigen_soc/sorad.f90

@@ -46,7 +46,11 @@ CONTAINS
     IF (atoms%jri(ntyp)>atoms%jmtd)  CALL juDFT_error("atoms%jri(ntyp).GT.atoms%jmtd",calledby ="sorad")
     ALLOCATE ( p(atoms%jmtd,2),pd(atoms%jmtd,2),q(atoms%jmtd,2),plo(atoms%jmtd,2),fint(atoms%jmtd),&
          &   qlo(atoms%jmtd,2),plop(atoms%jmtd,2),qd(atoms%jmtd,2),v0(atoms%jmtd),vso(atoms%jmtd,2) )
-    !
+
+    p = 0.0 ; pd = 0.0 ; q = 0.0 ; plo = 0.0 ; fint = 0.0
+    qlo = 0.0 ; plop = 0.0 ; qd = 0.0 ; v0 = 0.0 ; vso = 0.0
+    
+
     DO l = 0,atoms%lmax(ntyp) 
 
        DO jspin = 1,input%jspins
@@ -130,6 +134,7 @@ CONTAINS
                 !+apw+lo
                 IF (atoms%l_dulo(ilo,ntyp).OR.atoms%ulo_der(ilo,ntyp).GE.1) THEN !  calculate energy derivative (of order atoms%ulo_der) at e
                    ALLOCATE (glo(atoms%jmtd,2),pqlo(atoms%jmtd,2),filo(atoms%jmtd,2))
+                   glo = 0.0 ; pqlo = 0.0 ; filo = 0.0
                    pqlo(1:atoms%jri(ntyp),1)=plo(1:atoms%jri(ntyp),jspin)
                    pqlo(1:atoms%jri(ntyp),2)=qlo(1:atoms%jri(ntyp),jspin)
                    i = atoms%ulo_der(ilo,ntyp)
@@ -190,6 +195,7 @@ CONTAINS
                       !+apw+lo
                       IF (atoms%l_dulo(ilo,ntyp).OR.atoms%ulo_der(ilo,ntyp).GE.1) THEN ! calculate orthogonal energy derivative at e
                          ALLOCATE (glo(atoms%jmtd,2),pqlo(atoms%jmtd,2),filo(atoms%jmtd,2))
+                         glo = 0.0 ; pqlo = 0.0 ; filo = 0.0
                          pqlo(1:atoms%jri(ntyp),1)=plop(1:atoms%jri(ntyp),jspin)
                          pqlo(1:atoms%jri(ntyp),2)=q(1:atoms%jri(ntyp),1)
                          i = atoms%ulo_der(ilo,ntyp)