Add test data versioning

[behrtam]

quote from the xgo readme

The benefit of all this is that nitpickers can see which test version a posted solution was written for and be spared confusion over why an old posted solution might not pass current tests.

I want to work on test generation in the coming days and I think versioning of test data might come in handy for this. Anyway it even makes sense without automatic test generation.

So, the question is how do we want to implement it.

xgo
They work with two variables in the program and in the test file which is checked by the test. If the test program or test data is changed it is manually incremented.

xruby
They also have the version in the program and in the test file, but also a dedicated .version file. If I understand that correctly the .version gets inserted into the *.rb, *.tt and *_cases.rb files by the generate script which also increments the number.
They only use versioning if they have a test generation script.

xpython
I think it makes sense to use versioning for both manual created and automatic generated tests as test data can and will change in both ways.
Right now I don't see any benefit in the dedicated version file.
It would also mean that the user has to create the version variable in every problem file where we don't have a stub.

[behrtam]

Maybe there will someday also be a version key in the canonical-data.json that might help with this task (see exercism/problem-specifications#336 for details).

[behrtam]

Versions are added gradually, so the next step is validation: exercism/discussions#133