Improve heuristics for chosing default number of scattering angles (-ntheta)

[GoogleCodeExporter]

Currently this heurestics is based on number of dipoles. This becomes 
especially inadequate when refining discretization of the fixed particle.

Much better alternative is to base it on size parameter that is also readily 
available in the code. Because size parameter determines the typical angular 
scale for variation of light scattering patterns (from diffraction-like 
arguments).

Original issue reported on code.google.com by yurkin on 8 Jun 2010 at 4:01

[GoogleCodeExporter]

Original comment by yurkin on 5 Feb 2014 at 11:49

• Added labels: Milestone-1.3