Check parasitic models against sky130 and gf180mcu data

[RTimothyEdwards]

The parasitic capacitance models in magic were recently updated to capture various important effects previously not handled, the main one being a halo area-of-effect for fringe (perimeter) capacitance. Previously magic modeled fringe capacitance as applying to an infinitesimally thin line along the boundary of a shape. The new model spreads the capacitance over an area extending outward from the boundary. These models likely need refining, namely in two ways:

1. A sine approximation is being used to capture the partial fringe contribution out to a specific distance, and also the effect of fringe shielding of nearby shapes. These may or may not be a good match to data (I believe the latter is a pretty good match, the former not so much). Models should be checked against a field equation solver like FasterCap.

2. The point of the sine approximation was to limit calculations to a halo area of effect. This halo is currently a single global variable, and almost certainly needs to be an array of values per layer type. This will require changes to the syntax of the tech file "extract" section.

Also:

3. Need to determine whether additional effects are important enough to present a significant difference between measured and calculated parasitic capacitance, and do include those effects in the models if needed.