diurnal cycle in TOP

Boolean l_trcdm2dc depends on boolean ln_dm2dc defined in namelist. It aims at averaging qsr for biogeochemical tracers if there is a diurnal cycle.

1. The name of the boolean "l_trcdm2dc" is confusing because it has the opposite function to the boolean ln_dm2dc. I suggest to call it "l_trcdc2dm" instead.
2. Currently, the daily mean on qsr for biogeochemical purposes only works if ln_dm2dc=T that is to say if there is an analytical diurnal cycle. It does not work if there is a natural diurnal cycle with hourly atmospherical forcings for example. I suggest to define a new boolean in namelist_top "ln_trcdc2dm" (instead of the internal boolean l_trcdc2dm) to specify if you want to average qsr or not for biogeochemical tracers ⇒ modify trcnam.F90 to read a new parameter
3. If the analytical diurnal cycle is used (ln_dm2dc=T and ln_trcdc2dm=T), we should keep the mean value in qsr_mean (no need to call subroutine trc_mean_qsr) ⇒ modifications in sbcblk.F90 and trcstp.F90