Quentin 03/09/2020: GFS bugfix: vertical interpolation...

Quentin 03/09/2020: GFS bugfix: vertical interpolation (ver_inter_to_mixed_grid.f90) from the GFS isobaric pressure grid was wrong.
Previously, the computation of the altitude of mass points XZMASS_LS (ver_prep_gribex.f90) was done from XZFLUX_LS computed from the orography PZS_LS (in change_gribex_var.f90) as for hybrid vertical coordinate.
However, the vertical grid of GFS is isobaric pressure (no orography). Thus, previously the first level of GFS data (at 1000hPa) was interpolated at the first grid level of MNH (orography + ZHAT) leading to strong biais of temperature, wind and so on near high topography. The corrections include :
- reading of all levels (31),
- add tyeOfFirstFixedSurface argument for SEARCH_FIELD routine to read correctly the orography and geopotential height from the GRIB,
- reading the geopotential height in read_all_data_grib_case.f90 for GFS and inject it in the case of vertical interpolation of an isobaric pressure grid in ver_prep_gribex_case.