(cherry picked from commit aa1fdae51201b3863955ce5a37ba9fe6deb81e86)

Philippe 22/11/2016: lfiz: use F2003 calls to get argument lists + modified LFI integers to support LFI_INT=8

(cherry picked from commit 5fd86cd509ffefe231cd009da8f94b92d2dbb246)

Philippe 05/03/2019: deallocate XMI and XSOLORG if already allocated in ini_modeln but size=0 with respect to commit 77ee6c7a (29/01/2019)

Soline Bielli : 01/03/2019: Sea salt update: the significant sea wave height now influences salt emission.
The sea salt is either set to a default value of 2meters or read in ECMWF Grib file (if the variable exists).
Now 5 modes are available (3 before) and the order in output is changed.
The large-scale sea wave height from ECMWF grib is 2D. HALO2 update of the ls variable in modeln is not yet adapted to 2D field (it is switch off in this version for this variable).

Quentin 01/03/2019: backup from commit 7d17f6da: the TINY function added on PURS lead to non-zero values to other variables such as convective precip. and snow variables such as T%TSNOW_ROOF%WSNOW in urban_snow_evol.F90 which lead to activate the snow cover 1layer model even if snow is not present

Quentin 25/02/2019: correction in chemistry: cancel of previous commit : no need of change BASIC.f90 related to MEGAN emission scheme (one BASIC.f90 by RELACS scheme. The default RELACS scheme in MNH is still the same)

Philippe 21/02/2019: bugfix: intent of read fields: OUT->INOUT to keep initial value if not found in file

Quentin 19/02/2019: adapt the chemistry-namelist for ICARTT KTEST to the new BASIC.f90 (commit of 14/02/2019) + add MNH_MEGAN in profile_mesonh.ihm

Philippe 19/02/2019: IO: simplification/restructuration/cleaning of open/close subroutines (to be continued)