Incoherence in dimension of variable stempdiag coming in to nitrogen_dynamics

[jgipsl]

Currently in the trunk (rev 6186), in the call to nitrogen_dynamics in stomate_main, stempdiag-tp00 (DIMENSION(kjpindex,nslm)) is given but nitrogen_dynamics is waiting for temp_sol with DIMENSION(npts).

Solution proposed by Nicolas Vuichard:
Change in nitrogen_dynamics to take as input variable stempdiag (DIMENSION(kjpindex,nslm) in K.
Then locally in nitrogen_dyanmics, make a new variable by integrating over soil layers as done in littercalc:

    !! 3.2.2 integrate over the nslm levels
    tsoil_decomp(:) = zero
    DO l = 1, nslm
       tsoil_decomp(:) = &
            tsoil_decomp(:) + tsoil(:,l) * rpc(:) * &
            ( EXP( -z_soil(l-1)/z_decomp ) - EXP( -z_soil(l)/z_decomp ) )
    ENDDO

Also transforme into deg C.

[jgipsl]

Done in the trunk [6235]. tsoil_decomp calculated in littercalc is used directly instead of recalculation.