attenuation on transversely isotropic model

[PengJX1023]

Hello, I found there is possible problem about the Lamé coefficients used to calculate attenuation of TI models.

when validating with AxiSEM3D, there is evident discrepency between waveforms for transversely isotropic model.

At the begining of surface wave, there is about 5s misfit in phase because of different Lamé coefficients used in calculating attenuation. I think it' more reasonable to use Voigt average Lamé coefficients than only use horizontal velocities to calculate $\lambda$ and $\mu$. The dealing method doesn't affect Constitutive relation, but do influence attenuation calculation later.

The command is in get_models.F90 Lines 173. the related codes in def_precomp_terms.f90 should also be adjusted.