I'm not going to say for certain that I have the correct solution yet, but I have found a test case that is setting an invalid expectation. With just 4 buckets, a 15 minute cooldown, and 30 minutes to locate the poison it is absolutely possible to do it with just 1 pig. Not guaranteed, mind you, as the pig could die in round 1, but the problem didn't ask for worst case (losing a pig each round), the problem asked for the best possible case (minimum number of pigs).

Therefore the expectation of an result of 2 pigs in this scenario is wrong, because it represents the worst case scenario.
This is further supported because at least one other test case assumes best case:
1000 15 60 = 5 pigs with all of them surviving every round. If you start with 5 and lose 1 each round, then by the final round you only have about a 66% chance of finding the single poisoned bucket. It's possible that the solution assumes that I lose 5 pigs and have a 6th that survives, but the problem description doesn't ask me how many pigs die, it asks me how many pigs I need.

So at least one of these tests is wrong, and possibly both are. Yes, 2 is the minimum number you need to allow for pigs dying with a 15 minute cooldown and 30 minutes alloted, but that's also not the question that was asked for this problem.

Could we be more explicit with the description of the problem, or else fix the test cases to match the existing description? I'm going to see if it works out assuming that I'm counting fewest possible dead pigs now, as that seems to be what the tests are looking for.

Thanks.