To be fair, you seemed to know what you are doing, but I'd rather state the obvious than have you gas yourself... before we find out what the problem is anyway But then I'm assuming you also have a volt meter then, which means you could connect it to the switched live (you might be able to use the 'test socket' terminals on the boiler) and will make a diagnosis easier.
I'm clutching at straws too, but we're thinking along similar lines. The pump is switched by the boiler and the over-run stat could power it with the system controls not feeding the switched live. But that wouldn't explain it if the gas valve is also opening. Unless somehow we have 1,2, and 3 all connected together somehow.
I suppose at this point I'd want to confirm whether, during this weird erratic behaviour, the pump is receiving power, whether the boiler is firing, and whether the gas valve is receiving power as certain hypotheses could then be ruled out. I find that sometimes in these situations, careful observation reveals it all should have been obvious all along.
I'm clutching at straws too, but we're thinking along similar lines. The pump is switched by the boiler and the over-run stat could power it with the system controls not feeding the switched live. But that wouldn't explain it if the gas valve is also opening. Unless somehow we have 1,2, and 3 all connected together somehow.
I suppose at this point I'd want to confirm whether, during this weird erratic behaviour, the pump is receiving power, whether the boiler is firing, and whether the gas valve is receiving power as certain hypotheses could then be ruled out. I find that sometimes in these situations, careful observation reveals it all should have been obvious all along.