uses numpy inf instead of LARGE_NUMBER

[tylerflex]

Previously, we defined td.inf to point to a LARGE_NUMBER defined in components/constants.py. This got the job done but had a few downsides:

• Was not explicit in the data files whether a number was just large or was supposed to be infinite.
• Needed to still be careful we got expected behavior if comparing this with another number that happened to be large (like frequencies in Hz).

This PR sets td.inf = np.inf, which seems to write to json fine.
The visualizations also render correctly.

I had to add a @staticmenthod in Geometry to evaluate the infinity values as LARGE_NUMBER in before setting axes limits as matplotlib could not handle this.

Everything else should be good to go.