rna-transcription: Add test case for invalid input

[MazeChaZer]

The return type of Maybe String already anticipates error handling, but I've seen many solutions on exercism.io that would just throw an exception on invalid input. This test should point people in the right direction.

[petertseng]

In exercism/problem-specifications#1027 three error cases were removed by the result of a coin flip. Now it is left up to the decision of individual tracks whether to add them back. That is really unfortunate because I really don't care to make such a decision, but I'll resolve the problem by simply saying if someone will do the work I'll merge it.

If the Haskell track is wanting to test invalid invalid inputs, I would suggest using those exact three case that were removed in exercism/problem-specifications#1027 and the justification for keeping them can just be that the justification for removing them was not strong enough.

[MazeChaZer]

I updated the file with the prior examples from the specification.

[petertseng]

this will be great, thank you.

[petertseng]

confirmed that this file is now the same as it was BEFORE 07c7426 (#634)

[petertseng]

I'm going to use something like this

rna-transcription 1.2.0.7: Revert "rna 1.1.0.5: remove Nothing tests"

This reverts commit 07c7426.

In exercism/problem-specifications#1027 three
error cases were removed by the result of a coin flip. Since that is
insufficient reason to remove them from the Haskell track, the Haskell
track chooses to keep them, and needs no further justification to do so.