It can also be a 0 :'The cells of the sudoku board may also contain 0's, which will represent empty cells.'

I did notice a bit unusual display of test cases, but I did not see it's that confusing. It's probably of no help for you anymore, but I rased appropriate issue above.

I just solved the kata in Scala and encountered no problems. I am going to close the issue here as not a kata bug.
If you want us to take a look at your solution, post it here (remember about markdown code formatting and spoiler flag) so we could take a look at it.

Interleaved tests output can be a sign that test runner executes tests in parallel, but AFAIK there were some countermeasures taken against such behavior. I will verify if it's indeed the case. This can also cause some problems if your solution has some state, but you said it does not.

right, it did sound familiar but I didn't check.

That's basically what I meant "My solution works "on my machine", but fails when I submit it to CodeWars. Why?" to be, in particular the point about state being carried over between invocations of solution. I did not think of all scenarios apparently, and sure, we can add them there.

@hobovsky: we'll need to add a point about "my code doesn't output the same on cw and my local environment"

-> global vars / language versions / mutation of the input / hidden memoization (python specific, with mutable default arguments) / ...?

I haven't solved it in Scala so it is possible there's some issue there, however, for such a basic test case, I kinda doubt it. Please read this point of the troubleshooting FAQ, especially bullet point no. 4, and see if it helps. I will check Scala version later today, but still I am pretty convinced the error is somewhere in your solution. If you find it, remember to mark the issue as resolved.