I would not make baclin that small. It is typically 8 (or many more) batrop time steps.
Your problem may be the bathymetry:
reading bathymetry file from regional.depth.[ab]
bathymetery from 30-second GEBCO_08 20091120 global dataset, new 11x11 average
i/jdm = 225 160
plon,plat range = 47.50000 58.70000 23.49952 30.57386
Filled N and S edges for CICE
min,max depth = 0.04364 3233.82178
The minimum should be something like 5m (we have used 2 m in the Gulf of Mexico). There is no need to change the coastline, just clip all depths less than 5 m to 5 m. I would clip and then smooth 1x. See the attached scripts. Note that if you leave the coastline the same, the only new files you should need are relax (and depth obviously).
Alan.