The issue of stagnation comes up particularly when you try to get a higher solar yield - in summer you have more panels than you need so often by the afternoon the cylinder is hot, the pump is off and the panels at 170C or so. I understand it is best to have systems that are arranged to get the working fluid out of the panels easily eg by a bottom connection for a flat panel, but how well do evacuated tubes work in this case, and are the heat-pipe ones better?
The issue is the volume of pipe/manifold that is directly heated, which dictates the volume of gas that is formed in a well-designed system. Generally heat pipe ETs have least volume, then through-flow ETs and then flat plates. Good system design involves leading the pipes downwards from the manifold, among other things, so that gas does not preferentially rise into the pipe.
Also, do auto air valves above the panel help things? I imagine selctive venting of gases might upset the composition of the heat transfer fluid, and so better to design without, if possible.
Auto-venting is a bad idea. The liquid that is lost has to be replaced and that may well upset the composition as you say and refilling usually involves manual intervention. A pressure relief valve is essential, but the system should be designed with an adequately large expansion vessel so that stagnation does not trigger sufficient pressure rise to cause venting.
An alternate solution is to add a heat dump mechanism - typically a radiator - through which flow is diverted to dissipate heat when the store is full. Sometimes the central heating circuit is used. Strategies are still required to deal with stagnation and overpressure should they occur.
Is there any long term evidence on the impact of high temperature stagnation on the reliability of different solar thermal systems?
I expect there is but I can't cite it. You may well get an answer to this question on the Navitron forum if not here.