Arrows indicate direction of voltage for "more dimming" with current controlled by voltage drop across twin R's 2R2 from 5.2V. It should be controlled by remote Brilliance on screen and Light Sensor. But this is a more direct way .
Monitor voltage on test points 1094, 1040, 1038 if you need help choosing an R value and report back.
Then contrast can be increased on display to adjust black level as full black.
There are better ways, but more complex.
1) LED drivers work differently, not with power but with current. I expect that the old 33.6 W driver outputs 1 A (of current) and up to about 36 V. You have to use it with LEDs which can handle 1 A or more. If the LED is 1 A, 20 W then the driver will supply 20 W to that LED since the current is limited to 1A. Both LEDs and drivers should operate below their maximum rating (of course).
2) Correct, halving the current to 500 mA will indeed lower the power. Now that 1 A 20 W LED would run at 0.5 A and use about half the power. That does not mean the light output is also halved, many LEDs are more efficient at lower currents so you might get 60% of the light instead of the 50% you'd expect. Powering a 20 W LED at 10 W will increase its lifetime and reliability a lot assuming you keep it cool that is. Heat is a LEDs enemy !
3) LEDs need a certain voltage, in this case much more then that 12 V. So your "powerfull" 12 V 5 A supply cannot deliver the power to the LEDs. The voltage is too low for the LEDs. This supply is constant voltage (12 V), for LEDs you need constant current. Constant voltage supplies are very much unsuitable for most LEDs. Only 12 V LED strings (for decoration) can use a 12 V supply.
4) Depends on the LEDs. Most LEDs modules I've seen are 36 V nominal, that's about 10 LED chips in series. Don't pay too much attention to the maximum voltage, the current is more important. I expect that 44 V max is often enough.
5) Depends on the driver. But controlling the voltage with LEDs is impractical, like mentioned above, LEDs need to be driven by a current, the voltage is largely irrelevant. A LED can be dimmed by controlling the current. But a nicer more controlled way of dimming a LED is done by using PWM (Pulse Width Modulation). It means switching the LED on and off quickly, for example 400 times per second. Human eyes are too slow to see that but some animals can see that. I would avoid a PWM based controller for an aquarium.
6) see 5) as I said, forget about the voltage, it is the current (which is measured in Ampere) which matters.
If you'd use a voltage to dim a LED, the light output would vary over temperature, age and whatnot of the LED. When using current control this is much less an issue, light output is much more predictable and controllable. This has to do with the Voltage/current behaviour of a LED.
Best Answer
I don't see much contradiction. Inrush current is usually an exponentially decaying spike. The "7A" nameplate value is certainly the peak current. The "300ms" is likely a nameplate for spike duration, probably defined at 10% level, so the total energy is much lower than the bold estimation. Here are "definitions" from Murata:
where the duration looks like is defined at 0% level :-)
There is a line-up of technology and tutorials that deals with inrush current measurement, like Keysight
Let me guess some numbers from the Philips datasheet.
If the input has 7A at 240 V peak, the ESR looks like about 35 Ω. If the 300 ms is defined as RC constant, then the capacitor might be (35 * C = 0.3) C =8,500 uF, which sounds too high. So the 300 ms is likely defined differently in the area of industrial lighting.