test: add a fail-test for escaped non Unicode Scalar Values

[eugenesvk]

Closes #451

[zkat]

We should probably have separate failing cases for each thing we intend to test, otherwise a single (correct) failure might obscure failures for other items

[eugenesvk]

Thought about that, but that's solved via parameterized tests, then can properly test that each \u codepoint in the range fails, otherwise didn't want to create hundreds of separate files, and without that it's no help separating 5 codepoints currently in the text out

[zkat]

I would prefer not to complicate the test suite right now. 4-6 tests at the appropriate boundaries should be “good enough”