- modd_budget : create TBUDGETCONF_t type and LBUDGET_* as POINTER
- ice_adjust_elec, lima_adjust_split : comment CALL CONDENSATION for the moment
- ini_rain_ice_elec : move allocation of variables to rain_ice_allocate (except XGAMINC_RIM3 not coded yet in modd_rain_ice_param (==>TODO for elec ?)
- mode_fill_dimphyexn : missing USE MODE_ll for GET_INDICE_ll
- resolved_cloud : use of CST_t + add new args for ICE_ADJUST (YLDIMPHYEX,CST, RAIN_ICE_PARAM,NEB, TBUCONF + LMFCONV)

Is the call of FILL_DIMPHYEX in resolved_cloud correct ?

Fields not bit_repro : THS_CLD, RVS_CLD, RCS_CLD, CLDFR

But is it really not surprising ?

- modi_condensation : remove MODD_SPP_TYPE (not used)
- condensation : fill HALO (NaN present) of PRC,PRV,PRI (OUT) with IN values (should be only on HALO instead of full copy) ==> on GPU branch, NaN should not trigg an exception in ice_adjust on the test of ZW1< 0.0  (SR)
+ PT computation on KIB:KIE only (because ZFRAC is defined on physical point only) in the CALL COMPUTE_FRAC_ICE
- modi/ice_adjust, condensation : move the TBUDGETS/KBUDGETS args (non optional possible) before the optionnal args
- mode_icecloud : modi_tiwmx was removed and the QSATMX_TAB is now in MODE_QSATMW_TAB
- lima_adjust_split, : SIGQSAT is now 2D