Divid by zero possibility in CBCK12

[langevin-usgs]

The revised code for calculating K for anisotropic cases could have a divide by zero if cl1 or cl2 is specified by the user as zero. I suggest that we construct the denominator here in pieces as ANUM or HYMEAN could possibly be zero if their lengths (EL1 or EL2) are zero.

        ACA = COS(ANGLE) * COS(ANGLE)
        ASA = SIN(ANGLE) * SIN(ANGLE)
        ANUM = 1.0 / (ACA/ANUM + ASA/HYMEAN)