Node synthesis

[tomalrussell]

What's the problem this feature will solve?
The core functions of the package create edges between a provided set of nodes. Examples in the documentation demonstrate uniform random sampling within a 2D bounding box to create a set of nodes. In general, to create complete synthetic spatial networks, there is a need to synthesise nodes with locations.

Describe the solution you'd like
One or more methods to generate sets of nodes to seed complete network creation:

• as in the docs, uniform random sampling within a bounding box

Perhaps also:

• sampling masked by a polygon (e.g. to sample points over land)
• sampling weighted by a distribution (e.g. according to population density)
• sampling uniformly on the sphere

[wang-boyu]

I agree that creating synthetic geometries can be useful, but I would prefer to consider it as out of scope at this point, since currently pysgn focuses on generating synthetic networks out of existing data. Plus, in #24 I have replaced the synthetic points with data from geodatasets, with only two exceptions:

1. In section Example: Geospatial Watts-Strogatz Network, a set of grid points is used to visually illustrate the concepts of local connections and rewiring;
2. In the last Performance section, random points with varying sizes are created to test time and space complexities.

As such there aren't many places in this codebase to use methods of creating synthetic geometries.