Added optional ozone tracer with chemistry parameterized by Daniel
Cariolle. This tracer is passive: it has no influence on the rest of
the simulation.
Added variable "zmasse" in file "histrac.nc". Corrected long name of
variable "pplay" in "histrac.nc". Changed name of variable "t" to "T"
in "histrac.nc", corrected long name and unit.
In "phytrac", moved definition of "zmasse" toward the beginning of the
procedure, so that "zmasse" can be given as argument to
"traclmdz". Also added arguments "julien", "gmtime" and "xlon" to
"traclmdz". The four additional arguments are required for the ozone
tracer.
In module "traclmdz_mod", factorized declaration "implicit none" that
was in each procedure. There is now an equivalent single declaration
at the module level.
In procedure "traclmdz", removed variable "delp". Use "zmasse * rg"
instead since we now have "zmasse" as an argument.
Tests. Compilations on Brodie only, with optimization options "-debug"
and "-dev", parallelization options "none", "mpi", "omp" and
"mpi_omp", this revision and revision 1373. Run cases with and without
ozone tracer, 1 and 2 MPI processes, 1 and 2 OpenMP
threads. Comparisons of all cases are ok, except for strange
variations in variables "d_tr_cl_RN" and "d_tr_cl_PB" of file
"histrac.nc", variables "RN" and "PB" of "restart.nc", variables
"trs_RN" and "trs_PB" of "restartphy.nc". Relative variations of these
variables between cases are of order 1e-7 or less, after one day of
simulation.
