* NLFINPRAR in TFILEDATA type is now of KIND=LFI_INT
* use TPFILE as argument instead of its fields for FMOPEN_ll and FMCLOSE_ll

(cherry picked from commit 6c172b3f)

(cherry picked from commit 4691b211)

Philippe 31/10/2017: IO: pass TPFILE to IO_WRITE_FIELD_LFI_xxx calls and use it to retrieve the logical unit + correct a bug in the integer kind of the LU

Philippe 30/10/2017: IO: DIAG: minor bug correction: renamed field PABST in PRES (when written in hPa)

* replace FMREAD/FMWRIT by IO_READ/WRITE_FIELD
* modify SET_CONFIO_ll call in mnh2lpdm.f90
* use TZPGDFILE for read_chem_data_netcdf_case.f90
* declare TZATMFILE when necessary

Phasage 5.3.1 -> 54X-dev-IO
Warning: not yet fully OK (compilation fails and some old IO subroutines are used)

# Conflicts:
#	MY_RUN/KTEST/011_KW78CHEM/002_mesonh/run_mesonh_xyz
#	src/LIB/SURCOUCHE/src/modd_io.f90
#	src/LIB/SURCOUCHE/src/mode_fm.f90
#	src/LIB/SURCOUCHE/src/mode_netcdf.f90
#	src/MNH/advection_metsv.f90
#	src/MNH/c2r2_adjust.f90
#	src/MNH/call_rttov11.f90
#	src/MNH/call_rttov8.f90
#	src/MNH/check_zhat.f90
#	src/MNH/check_zs.f90
#	src/MNH/compare_dad.f90
#	src/MNH/compute_r00.f90
#	src/MNH/default_desfmn.f90
#	src/MNH/diag.f90
#	src/MNH/free_atm_profile.f90
#	src/MNH/ice_adjust.f90
#	src/MNH/ice_adjust_elec.f90
#	src/MNH/ini_aircraft_balloon.f90
#	src/MNH/ini_cpl.f90
#	src/MNH/ini_deep_convection.f90
#	src/MNH/ini_lb.f90
#	src/MNH/ini_ls.f90
#	src/MNH/ini_modeln.f90
#	src/MNH/ini_param_elec.f90
#	src/MNH/ini_prog_var.f90
#	src/MNH/ini_radiations.f90
#	src/MNH/ini_radiations_ecmwf.f90
#	src/MNH/ini_segn.f90
#	src/MNH/ini_size_spawn.f90
#	src/MNH/ini_sizen.f90
#	src/MNH/ini_spectren.f90
#	src/MNH/ini_surf_rad.f90
#	src/MNH/init_for_convlfi.f90
#	src/MNH/khko_notadjust.f90
#	src/MNH/lima_adjust.f90
#	src/MNH/lima_warm.f90
#	src/MNH/ls_coupling.f90
#	src/MNH/menu_diachro.f90
#	src/MNH/mnh2lpdm.f90
#	src/MNH/mnh2lpdm_ech.f90
#	src/MNH/mnh2lpdm_ini.f90
#	src/MNH/mnhread_zs_dummyn.f90
#	src/MNH/mnhwrite_zs_dummyn.f90
#	src/MNH/modd_budget.f90
#	src/MNH/modd_fmout.f90
#	src/MNH/modeln.f90
#	src/MNH/modn_backup.f90
#	src/MNH/paspol.f90
#	src/MNH/phys_paramn.f90
#	src/MNH/prandtl.f90
#	src/MNH/prep_pgd.f90
#	src/MNH/prep_real_case.f90
#	src/MNH/radiations.f90
#	src/MNH/rain_c2r2_khko.f90
#	src/MNH/rain_ice_elec.f90
#	src/MNH/read_dummy_gr_fieldn.f90
#	src/MNH/read_field.f90
#	src/MNH/read_grid_time_mesonh_case.f90
#	src/MNH/read_hgrid.f90
#	src/MNH/read_hgridn.f90
#	src/MNH/read_precip_field.f90
#	src/MNH/read_surf_mnh.f90
#	src/MNH/resolved_cloud.f90
#	src/MNH/set_grid.f90
#	src/MNH/set_ref.f90
#	src/MNH/set_subdomain.f90
#	src/MNH/shallow_mf_pack.f90
#	src/MNH/spawn_field2.f90
#	src/MNH/tke_eps_sources.f90
#	src/MNH/turb.f90
#	src/MNH/turb_cloud_index.f90
#	src/MNH/turb_hor_dyn_corr.f90
#	src/MNH/turb_hor_sv_flux.f90
#	src/MNH/turb_hor_thermo_corr.f90
#	src/MNH/turb_hor_thermo_flux.f90
#	src/MNH/turb_hor_uv.f90
#	src/MNH/turb_hor_uw.f90
#	src/MNH/turb_hor_vw.f90
#	src/MNH/turb_ver.f90
#	src/MNH/turb_ver_dyn_flux.f90
#	src/MNH/turb_ver_sv_flux.f90
#	src/MNH/turb_ver_thermo_corr.f90
#	src/MNH/turb_ver_thermo_flux.f90
#	src/MNH/ver_thermo.f90
#	src/MNH/write_balloonn.f90
#	src/MNH/write_budget.f90
#	src/MNH/write_diachro.f90
#	src/MNH/write_dummy_gr_fieldn.f90
#	src/MNH/write_lbn.f90
#	src/MNH/write_lfifm1_for_diag.f90
#	src/MNH/write_lfifm1_for_diag_supp.f90
#	src/MNH/write_lfifmn_fordiachron.f90
#	src/MNH/write_lfin.f90
#	src/MNH/write_phys_param.f90
#	src/MNH/write_zsmt.f90
#	src/MNH/zoom_pgd.f90
#	src/MNH/zsmt_pgd.f90
#	src/Makefile
#	src/Makefile.MESONH.mk
#	src/SURFEX/zoom_pgd_cover.F90

These subroutines were not using the correct buffer if CFORMAT='LFICDF4'

Philippe: 24/10/2017: workaround in PREP_PGD due to NPROC set earlier than before (commit 86f22a35)

Philippe: IO: allocate ZFIELDP/IFIELDP with zero size to remove alerts with Valgrind (not really necessary)

Philippe 24/10/2017: IO: bug correction: IO_READ_FIELD_BYFIELD_L1: size of message in MPI_Bcast was wrong (1 instead of size(ofield))

Merge 5.3.1 into 54X

Merge branch 'MNH-53-branch' of mesonh2.aero.obs-mip.fr:/gitrepos/MNH-git_open_source-lfs into MNH-53-branch

Philippe 17/10/2017: problem with SNGL intrinsics on BlueGene/Q => transform it by REAL in preprocessing

+ FMOPEN_ll/FMCLOSE_ll are now private
+ cleaning of comments