Distributions of GalSim object input parameters

[barnabytprowe]

For generating multi-object simulations we may want to give the user the option of specifiying distributions of parameter values from which to sample. This is the issue for discussing how to do this, and for what distributions.

[rmjarvis]

This is relevant to the discussion on #101 about what kind of parser we want.

e.g. I think it might be nice to be able to have configuration lines look something like

centroid_slop = gauss2d(sigma = 0.1)

psf_sigma_func = poly2d(0.7, 0.01, 0.03)  # (a,b,c) for a + b*x + c*y
psf = gaussian(sigma = psf_sigma_func)

which would probably push us towards Jim's eval idea for the configuration reading, since this is probably a lot closer to being possible as direct python code, rather than through something like ConfigParser.

[barnabytprowe]

I agree Mike, I've discussed this with Rachel here and we both agree that the potential power of being able to use the eval() function is worth exploring, and makes full use of the power of the Python platform.

[barnabytprowe]

I'll see if I can make a couple of examples...

[barnabytprowe]

Pull request #135 relevant to this issue.