First thing to do was to give some visual indication of the heater status. I decided to 'borrow' the headlight full beam warning light, which although fitted, hasn't been wired in yet.
Next I tried too reproduce the problem, this time by driving to the Post Office, sending off some build-DVDs to a new Superspec owner, then driving back. and to add another factor I stopped for another couple of minutes in a layby.
Here is the RPM display. So the profile was 7 mins to the Post Office, 2 mins with the engine off, 2 minutes driving, 2 mins with the engine off then 3 mins home.
And here was the lambda trace. So I guess the good news is that the problem was not a one-off but is reproducible, so that will make testing easier.
We can see that while sitting at the Post Office the lambda sensor went to it's default 0.45V although the ECU still showed it's status as being on-line (it's a bit odd but we can only assume that was a decision by the Rover programmers). But when I started the engine at the 9 minute point, it didn't react enough to the exhaust gases and at the 10 min point the ECU have up and switched it out of the circuit. Even on the next engine start at the 15 min point it still isn't really reacting properly.
I need to do another test run tomorrow, but this time switching the ignition off between sessions (I had to leave it on to keep the data flow going), to see if rebooting the ECU helps.
However, the really important thing is that for all 3 of the engine starts, the warning light illuminated so we can rule out the relay as the problem. So I guess the next thing to do is try a replacement sensor.
I have a couple of spare ones but unfortunately, last time I rebuilt the manifold/pipe join I stupidly positioned the sensor so there isn't enough clearance with the body to get it out without taking the exhaust pipe off.
So I'll leave that test until the next (inevitable) time when I have to remake the joint again.