Add bowl computation and use in bining

[eguil]

• The bowl is the deepest mixed layer reached over the period of study. It separates the upper ocean (in contact with surface fluxes) from the deeper adiabatic ocean.
• It is a fixed surface in time: bowl(i,j).
• Mixed layer criteria should be density based
• Added comment: looking at CMIP5 output, the right variable would be 'mlotst' but very few models provided it. Alternatively, one could use 'omlmax' (based on ML scheme) or derive the MLD from monthly T and S and a 0.01 delta density criterion (to be checked against mlotst when available).

[durack1]

A possible diagnostic could be % of time that a latitudinal density point exists - so in Durack & Wijffels (2010) the assumption was made on a monthly climatology, so if an isopycnal exists 12 months of the year this is the limit of the bowl

[durack1]

And here is a published plot from Durack & Wijffels (2010; http://doi.org/10.1175/2010JCLI3377.1)

[eguil]

Persistence computation added (in %): map + zonal mean per basin (files _persistent) Depth / temp/ salinity of shallowest persistence ocean added (zonal mean)