Add 0.2K temperature mixed layer depthdiagnostic

[jwhile]

Add a new MLD diagnostic to NEMO. The diagnostic will be based on a 0.2K change in temperature from the surface - this is the same definition as used by deBoyer ​http://www.ifremer.fr/cerweb/deboyer/mld/home.php.

This will add an additional mixed layer depth diagnostic NEMO outputs, with the output controlled by a namelist. This criterion is used by the Met Office's data assimilation system in shelf seas.

This is a simple update to add a new diagnostic and it will not change any physics. Consequently the only tests I propose are:

1. Run the stem tests.
2. Verify the diagnostic by running the model for a few days and verifying that the MLD is output. The values obtained can then be compared to the deBoyer MLD climatology to make sure they are sensible.

Paolo are you happy with these tests?

[smasson]

???
I am afraid this is already existing...
see diahth.F90 and the following variables:

      CALL iom_put( "mld_dt02", zabs2        )   ! MLD abs(delta t) - 0.2
      CALL iom_put( "topthdep", ztm2         )   ! T(10) - 0.2
      CALL iom_put( "mldr10_3", zrho10_3     )   ! MLD delta rho(10m) = 0.03
      CALL iom_put( "pycndep" , zpycn        )   ! MLD delta rho equi. delta T(10m) = 0.2
      CALL iom_put( "BLT"     , ztm2 - zpycn )   ! Barrier Layer Thickness
      CALL iom_put( "tinv"    , ztinv        )   ! max. temp. inv. (t10 ref) 
      CALL iom_put( "depti"   , zdepinv      )   ! depth of max. temp. inv. (t10 ref) 

I don't really see the need to use a namelist parameter since we have the xml interface of xios...

[lovato]

The diagnostic is already included in both 3.6_stable and trunk codes (see NEMO/OPA_SRC/DIA/diahth.F90)
and its output can be easily handled with XIOS.