Something was not mentioned here: F36T8s have 9V preheat cathodes..that's also why they don't work well on Rs in the US

The heating is necessary only for the start. If it is a full power ballast, these lamps need no heating supply whatsoever, the cathode fall is sufficient.
However the US RS ballasts deliver too low voltage to ignite the F36T8 (F40 suffices with about 300V peak, so 210Vrms, while theF36 needs more than 400V), that is the main problem with RS. On preheat or most electronic they should work fine.

But hey, wouldnt that mean that the high voltage cathode would be overpowered and overheat on European RS ballast ? (which is roughly a current source when there is no arc)

Cause that is not the case, it is barely visible dim red

This is getting very confusing.

But hey, wouldnt that mean that the high voltage cathode would be overpowered and overheat on European RS ballast ?
Cause that is not the case, it is barely visible dim red

The filaments are designed to be preheated by about 1.5x the nominal arc current (that is, what the preheat ballasts usually deliver into a short circuit; it is assumed to be time limited, the "unlimited" rating is then about 1.3x the ballast rating). That yields the 9V rating during preheat phase.
In European RS ballasts (a.k.a. SRS) the heating current disappears once the arc ignites (it is the way, how the current phase shifts work in that circuit). But due to the topology, the preheating current is only what flows through the capacitor, so a bit less than the rated arc current. It is sufficient to reach at least some emission, so to allow an ignition, but then the filaments are supposed to completely warm up from the arc by itself. And that is, why you see the cathodes only dim red - the current is just not sufficient for a full warmup.

When the lamp is burning, only part of the filament (in a preheat circuit, where the current comes just from one side) is biased by the current and that current is what the ballast feeds. But that means way less than 50% of the power the filament needs to remain at the emission temperature, the rest comes from the cathode fall (in fact the energy of the ions hitting the cathode). In SRS the current emerges practically from both sides of the filament, so the only heat source is the arc cathode fall itself.
Normally the cathode fall responds to the cathode emission, hence the cathode temperature and that mechanism maintains the temperature exactly on the desired level (lower emission -> higher fall -> faster ions -> higher power dissipation -> the cathode heats up). The thing is then designed so, the resulting cathode fall is low enough to prevent sputtering, but still high enough so that mechanism has enough room to correct variations (so the filament would not overheat).
And this mechanism then settles on higher cathode fall, when the arc current becomes smaller.
And that will happen when a 0.43A rated tube gets operated from a 0.25A ballast without sufficient auxiliary electrode heating.

On the other hand that difference is not that huge (there is yet another effect: The heat is being lost by the emitted electrons not returning back, so higher current means more heat removed from the cathode), so the cathode drop will increase from about 15V to less than 25V or so, what is still not that excessive as e.g. cold starts (there we are talking about above 100V or so).

This is getting very confusing.

The matter is not really simple, there are many effects, each of them influencing the lamp wear in different direction. So just telling which one is dominant and how much is not that straight forward.
Normally lamp makers make sure to design the lamps so, the rated conditions correspond to the optimum (where the life is the longest), what they were able to reach given the conditions they have (manufacturing capability, engineering excellence, even the necessity to not infringe someone else patents,...). So in 99.9% any deviation from the rated operating conditions means worsening the lamp wear and/or performance. How much and which mechanism exactly, no one really knows...

Then the low resistance cathode of 40W would have even less power delivered to it than the 36W when there is no arc. How is that sufficient ?

Then the low resistance cathode of 40W would have even less power delivered to it than the 36W when there is no arc. How is that sufficient ?

It is meant to just lower the ignition voltage of the arc so the mains voltage became sufficient to ignite it. Then the full warmup relies on the arc itself.
The idea of fully warming up the electrodes before ignition to actually eliminate the starting wear and so allow for frequent switching came way later, not earlier than 80's or so. Till then the electrode preheating was really meant just as a method to ease the starting, nothing else.
And for the starting the cathode does not have to actually emit the electrons, way sufficient is to just reduce the electron liberation energy, so the electrical field needed to rip them off the cathode become attainable with just the mains voltage or a limited inductive kick.

But RS do very good job at keeping lamp life, so what is there is sufficient

But RS do very good job at keeping lamp life, so what is there is sufficient

Indeed, even when it is not at the full emission temperature, just warming it up big part of the temperature difference means the residual time of a cold cathode discharge is greatly reduced. Plus even the partial liberation energy reduction means the cold cathode voltage is already way lower than with really cold cathodes, so even with the shorter glow discharge, the ion's energy is lower as well.
In that respect the glowbottle preheats uses to be worse...


The design of the modern programmed start ballasts (the fluoractor based electronic starters belong there as well) really aims to warming the electrodes fully, so really completely eliminating the starting related wear, so go way further with that.

What is in this "Fluoractor" anyway ? Can a "general purpose" thyristor with appropriate ratings be used instead ?

What is in this "Fluoractor" anyway ? Can a "general purpose" thyristor with appropriate ratings be used instead ?

It is a thyristor  with the structure tuned so, the gate circuit may control the holding current, specifically designed for the use in these electronic starters.
So normally it is triggered via a Zener+resistor, when the voltage exceeds the programmed trigger voltage (200V or so for the F36T8 or so). Then for the first 2 seconds or so the gate is left floating, so the thyristor continues conducting till the current really drops to zero, so there is just the preheat current flows.
After the preheat timer expires, the control circuit starts pulling the gate negative and by that the thyristor switches OFF before the current crosses zero, so by that creates an inductive kick on the ballast coil, so the ignition pulses. When the lamp gets ignited, thext halfcycle the voltage does not exceed the trigger voltage, so the thyristor remains OFF and does not interfere with the lamp.
If the lamp fails to ignite after about a second of pulsing, the timer shorts out the trigger circuit, so the thyristor then remains OFF.

These thyristors are quite useless for anything else, as the modifications intended to make the holding current controllable by the gate make the structure quite extremely sensitive to any fast di/dt transients, so even a few nF capacitor parallel to such starter kills it quite readily.

Can a substitute be made with general purpose parts ?

It will be simpler to make the complete starter based on the different parts (microcontroller + IGBT + some auxiliary "dirt" around).
The fluoractor allows the starter circuit to be really simple (6 diodes, 1 Zener, 4 resistors, two capacitors), but replacing the fluoractor would be really hard...
But they are still sold and are not that expensive, so it is way easier to buy a new one...