Improve Divide-By-Zero Handling

[Grochocinski]

• Introduced make_nonzero() function to make checking/correcting potential divide-by-zero errors easier
• Updated Strong and SPSA solvers and post_normalize() to utilize this new function
• Regenerated some CNTNEWS1 and IRONORECONT1 test results that had 0s instead of nans after post_normalize() behavior changed
• Regenerated test results for EXAMPLE1 with SPSA as the new divide-by-0 behavior slightly changed run results for SPSA. Other tests for SPSA were not affected and thus didn't need regenerated.

ASTRO-DF behavior still needs refactored (pending merge of PR #149 into dev)

[Grochocinski]

Update: after reviewing the divide-by-zero handling in ASTRO-DF, I no longer think any updates to it need to be made as previously thought