The Planning Commission report is correct. The combined Ganges - Meghna-Brahmaputra rivers all flow from out of the same delta (mostly in Bangladesh), and float as a lower salinity, more buoyant layer out into the Bay of Bengal, certainly for hundreds of kilometres, maybe for more than a thousand. Wind turbulence, sea currents, and diffusion eventually cause the low and high-salinity water to mix, but it's a slow process. Moreover, nearly all the sun's radiant heat is concentrated in the topmost 20 to 30 metres, thereby warming and expanding the water to give even more buoyancy. Global warming will amplify the density contrast still further, so I very much doubt if the proposals to divert some of the flood waters into dry land areas will do much to reverse the trend. But some oceanic modelling should elucidate the matter further.
This raises an interesting point. How closely does the current generation of Ocean-Atmosphere-Coupled Global Climatic Models capture the physics of thermal buoyancy in the Bay of Bengal? Any answers?