Fuzz input validators for trailing whitespace

[Matistjati]

At the time of writing, 16 Aug 2025, both https://open.kattis.com/problems/procrastination and https://open.kattis.com/problems/ekkiminnforseti have trailing whitespaces in their samples. This is very minor, but should IMO not happen.

This PR adds tests that check the input validator will reject this.

[Matistjati]

On another note, as far as I can tell, we use random without seeding. This seems like it could be extremely annoying in some edge-case. Do you see any problems in naively adding random.seed(42) to the top of verifyproblem.py @gkreitz ?

[gkreitz]

On another note, as far as I can tell, we use random without seeding. This seems like it could be extremely annoying in some edge-case. Do you see any problems in naively adding random.seed(42) to the top of verifyproblem.py @gkreitz ?

I'm not sure fixing a seed at the start of execution is great, as adding a new junk case (or any other use of random) can then change all values. I would prefer that for the cases where we want to fix a seed (probably all of them), we do random.Random(0).<call>, as is already done on https://github.com/Kattis/problemtools/blob/master/problemtools/verifyproblem.py#L958 . 0 could of course very well be replaced by 42.

For a couple of the cases, where we're adding whitespace, I also feel we could simply skip randomness. If we want the input validator to catch trailing whitespace, why are we adding 1-5 spaces, instead of always adding the smallest amount we want it to catch (1)? I would be surprised if there are common bugs where you properly forbid 1 trailing space, but catch 2...

[gkreitz]

While I added a comment about how we do random, I think this is a very good addition by itself, so I'll merge it now, and then we can do a separate PR if we want to adjust randomness per my comment above (explicit seeding in a few places, and just adding a single space/newline in a couple of cases).