It's not uncommon when people make new tests by copying old ones and forget to update names, resulting in test-suites with duplicate names. This makes -p not being able to differentiate between them.
I think it would be a nice QoL if tasty at least warned about such "mistakes".
I just spent an hour or so looking at the wrong test and not understanding why it doesn't work: surprise, it did; i was looking at the wrong one.