This resolves the bug trigged in th_r_from_thl_rt_1d.f90 at the LOG(ZT) when some HALO points has T values < 0.
TODO: the computation in compute_entr_detr should be done only on updraft points

(cherry picked from commit 37ea99b9)

Philippe 19/11/2021: fix: correct lateral boundaries fieldnames for LIMA variables (be careful: forward compatibility with files is lost)

Philippe 19/11/2021: change MASDEV version to 5.5.1 (necessary due to change in fileformat in next commit)

Juan, Quentin 26/10/2021 bugfix RESTART reproductibility : use real(kind=4) for computation of LB and LS fields + DRYMASS sources bugfix

Quentin 25/10/2021: bugfix RESTART reproductibility. In the case of no online chemistry (LORILAM=F) or LDUST/LSALT=T processes changing the aerosols (XAER)
at each time-step, XAER is computed once at the START segment in ini_ecmwf_radiations based on current pressure and not updated during MESONH except if LAERO_FT=T and CAER='TEGE'. At RESTART, the aerosols climatology is updated but the
pressure has changed. For reproductibility, one must use LAERO_FT=T and CAER='TEGE'. XAER_CLIM was not updated in the process done by CALL AEROZON in phys_param

Quentin 04/10/2021: bugfix of the previous commit. The diagnostic surfex variables were not written at any first output file for every segment (instead of not written at time = 0s i.e. KTCOUNT=1 )

Quentin 30/09/2021: bugfix RESTART reproductibility. The surfex online diagnostics (but included pseudo-pronostic variables such as TSRAD) are written at
a time step of +1 in advance with respect to the atmospheric variables which lead to wrong computation of fluxes (such as radiative) at the initial step of a RESTART

Quentin 09/08/2021: Python LIB, add fontsizes customization; possibility to change the prefix figure name; add addWhite colormap for either the higher (top) or the lower (bottom) value : whiteTop flag

Quentin 05/08/2021: Python LIB, read with Epygram (GRIB1/2, LFI and FA) function + allow keyfile to be any String

Revert "Jean-Pierre 02/08/2021: adding ERA5 analysis on pressure levels as coupling files and fixing bug when using GFS analysis"

This reverts commit c610e351.

Jean-Pierre 02/08/2021: adding ERA5 analysis on pressure levels as coupling files and fixing bug when using GFS analysis