For coolant to leave from the top of the rad, it would have to enter from the bottom of the rad.
If the tstat is closed, the lower rad hose is blocked, so suction at the top of the rad would have no effect. Like you're sucking on a straw with the end blocked.
If the tstat is open, the pump is sucking coolant through that bottom hose. So now there's someone else sucking on the other end of the straw, and they're better at it, and man that got weird fast.
So there's either a deadhead or suction at the bottom of the rad, and an unrestricted supply of coolant coming out of the engine at the top.
The datalog could be interesting, but leave the two-wire sensor alone and use the one-wire sensor for the dash (with the wire to the dash unplugged) instead. Two circuits trying to check resistance across one sensor could end badly, or at least give weird and unreliable data.
And yeah, at this rate I'm gonna end up with a full backup of Nico's service manual collection on my hard drive. Sometimes it's worth it, though--last night I stumbled on the "body sonic system" in the '84 ZX manual. Looks like it had kickers in the seats, hooked to the radio. Butt subs.
Oh, and RE convection. Some early cars and tractors did not have water pumps, and relied entirely on thermosiphon cooling. Heat rises in the block, cold falls in the radiator, coolant circulates. It even kinda works, mostly.