Testing for illegal utf-8 seq handling

[CraigKelly]

Added a test asset (illegal-chars.txt) from the original issue (#504) and a test to insure the changes in the PR (#506) work correctly.

[coveralls]

Coverage remained the same at 77.199% when pulling 1a87a18 on CraigKelly:illegal-test into a1ddbb7 on amperser:master.

[suchow]

Thanks! For reasons of copyright, I am concerned about including Denis Pelli's unpublished abstract as a test case within the Proselint source code. Can you reduce illegal-chars.txt to a minimal example that contains only the offending characters and perhaps, if you think it's useful, some text describing its contents (e.g., This file contains an invalid UTF-8 character...)? Then please rebase over master and I think we're ready to merge.

[coveralls]

Coverage decreased (-0.09%) to 77.379% when pulling ec38b00 on CraigKelly:illegal-test into e38328f on amperser:master.

[coveralls]

Coverage remained the same at 77.469% when pulling 6df1791 on CraigKelly:illegal-test into e38328f on amperser:master.

[CraigKelly]

Sorry about that - I should have modified the file from the start. I just got a little excited. I believe the test is now working and correctly reading/interpreting output.

[suchow]

Thanks!