Refactor xjson library

[litghost]

Some latent issues found during review from #1237:

• prjxray/xjson.py is poorly named. Should be renamed to something akin to sort_complex_data
• prjxray.xjson.extract_numbers won't handle mixed formatting well. Example given was extract_numbers("A100") > extract_numbers("100B"), which generates error:

In [1]: (100, 't') > ('t', 100)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-1-d06ca7865f0c> in <module>()
----> 1 (100, 't') > ('t', 100)

TypeError: '>' not supported between instances of 'int' and 'str'

• prjxray.xjson.extract_numbers and prjxray.xjson.sort could use some more unit tests, above example is a good one.