Hello,
We have no such limitation as 0.85, e.g. in rivers you could have elongated elements in the direction of the flow, and the Courant number in all directions could remain about the same. One recommendation is to avoid obtuse angles: it is known that a diffusion matrix may become antidiffusion on an obtuse angle. Though we care for this it gives a hint that obtuse angles in a triangle are weird.
Generally speaking I wil say that the quality will be the same everywhere if you have the same Courant number everywhere, hence for maritime applications a mesh size proportional to the square root of the depth.
With best regards,
Jean-Michel Hervouet