Draft: Resolve "TOP light scaling under seaice create anomalous blooms"

Closes #261

Testing the light-Chl coupling in MEDUSA alongside the other light attenuation scheme, It appears that the biology is doing abnormally well under sea-ice in the Arctic, except in time of perpetual night (i.e: November to mid-January). As soon as QSR is slightly rising, the phytoplankton blooms.

It seems this come from the light module trcopt, where the sea-ice impact on light is done this way :

      !     Photosynthetically Available Radiation (PAR)
      !     --------------------------------------------
      DO_2D( 0, 0, 0, 0 )
         zqsr_corr(ji,jj) = parsw(ji,jj) * qsr(ji,jj) / ( 1.-fr_i(ji,jj) + rtrn )
      END_2D
      !
      CALL trc_opt_par( kt, zqsr_corr, ze1, ze2, ze3 )
      !

The same mistake is done in PISCES' p4zopt

A simple solution would be to multiply instead of dividing by the ice fraction :

      DO_2D( 0, 0, 0, 0 )
         zqsr_corr(ji,jj) = parsw(ji,jj) * qsr(ji,jj) * ( 1.-fr_i(ji,jj) + rtrn )
      END_2D

see here a one year Hovmoller of the PAR. First plot using MEDUSA usual light scheme, and bellow using trcopt. NEMO-MEDUSA is running 1D here, in a location between Svalbard and Greenland. The x axis is the time of year in year (0.5 is june).

here the impact on Chl :

As we can see, there's definitely a problem there.