I was going to ask what part of that made you think that RAIDZ works the way you think it does, but it really doesn't matter--it doesn't work that way. As
@m0nkey_ already said, "the pool will be set-up using the maximum space available on the smallest drive." Data and parity are striped across all disks in the vdev (there is no dedicated parity disk, as you seem to think), and the effect of this is that the smallest disk in the vdev limits its capacity.
The equivalent of RAID6 is RAIDZ2, which is double parity
We all know that RAIDZ2 is double-parity, and we all know that RAID6 is double-parity. That doesn't mean that RAIDZ2 is RAID6, and indeed, it isn't. We insist on the distinction because it makes a difference. If you say you have a RAIDZ1 pool, we know that's a ZFS pool. If you say you have a RAID5 array, we can assume (and hope) that it's truly RAIDZ1, but it's also possible that you did something truly stupid like build a hardware RAID5 array and create a ZFS pool on top of that--and yes, we've seen this exact scenario; this isn't just a silly hypothetical.