Add test case(s) - specifically for "random" encyrpted data including NULs

[eengstrom]

Following #41, probably aught to add some test cases. In the specific case of #40, create an artificial test case that embedded NUL chars into the fake encrypted data so that we can test the insertion of that blob into the DB.

Simple repeated test cases with short unencrypted data (e.g. "foo") are less likely to trigger the generation of pseddo-random encrypted data including NULs. In my original issue (#39), I was hitting the issue more frequently with longer source data to-be-encrypted. Make sense, just from a probability perspective.