Fixes in diagnostic script for recipe_sea_surface_salinity

[sloosvel]

Describe the bug
@portegam noticed that the computations in compare_salinity.py used in recipe_sea_surface_salinity are not entirely appropriate to compute the metrics needed in CMUG D5.4. He proposes the following changes:

line 119 (old code) >      corr = pearsonr(data, reference, ('month_number', 'year'))
line 119 (new code) >    corr = abs(climate_statistics(reference, 'mean')-climate_statistics(data, 'mean’))

line 124 (old code) >      [0.25, 0.5, 0.75], ["0.25", "0.5", "0.75"], color="grey", size=7)
line 124 (new code) >      [3, 6, 9], [“3", “6", “9"], color="grey", size=7)

line 154 (new code) >     f”Correlation comparison in diferent regions for "
line 154 (new code) >     f”Absolute bias comparison in diferent regions for “

[sloosvel]

FYI @kserradell