Fix test for bad chunk data to allow for 0-sample deep chunks

[kdt3rd]

Signed-off-by: Kimball Thurston kdt3rd@gmail.com

[lgritz]

I don't understand this part of the code at all, but something about the structure of the change seems odd to me.

There are a number of tests that look like this

if (condition)
    error;

and this PR changes those to

if (condition || other_conditions)
    error;

Please help me understand how strictly expanding the set of circumstances that are considered an error will fix the bug wherein a correct file has been incorrectly flagged as an error? If you are just using || to catch more conditions, won't the same false positive still occur?

[kdt3rd]

well, it was if value <= 0 before, I've changed it to if value < 0 OR (equal 0 AND the other value != 0) instead of just the <= test

[lgritz]

Aha, of course. Two of the places you changed were < before and still after, and two changed from <= to <, and I did not notice that. OK, all looks good.

[kdt3rd]

yeah, I basically added other guards for the other values - the first one for the packed size we had added was due to a fuzz report, the other cases I added would be exposed by other fuzz tests that hadn't yet been triggered - pre-fixing failures

[lgritz]

Woo-hoo! OIIO bleeding edge tests passed last night!