So, I have an 18W fluorescent fixture with non-Philips preheat ballast that is labeled for 18W fluorescent lamp. But, when I tested it with watt meter, it says that it uses 39,7 watt. So my question is, does preheat ballast use more electricity than it's rated or because I'm using a non-Philips ballast that maybe it's for 40W tubes but labeled 18W and I'm over-ballasting it?

Inductive ballasts are relatively inefficient. It indeed likely uses about 40 watts. They can, however, specify a lower consumption. This has to do with the power factor. Back in the day, the standard analog electricity meter with the slowly spinning disk, did not pick up the consumption by inductive loads like old school ballasts, so you didn't pay for the power that you kind-of are usung. They would only 'see' the 18w of resistive load the lamp takes.

This power factor stuff is why all professional lighting installations with inductive ballasts include a capacitor to bring the power factor closer to normal. Because although the meter doesn't see the load, the generator in the power station most certainly does, and your wiring does too. We used to have a tanning bed with 10x 100w fluorescent in it, but no capacitors. Even though it only was specced at 1000w, it heated up its mains cord like crazy because apparently the designers thought 'Huh, about 5 amps, we can send that through the cheap thin 6 amp mains cord' while the actual power consumption was closer to 10 amps if you include the inductive load.

Modern digital meters *do* pick up on loads with weird power factors. It's al a very complicated story, but in the end it boils down to 'ignore it if you have an analog electricity meter, try and improve it if you have a digital meter and gotta pay for the extra power usage'.

The reason why inductive ballasts are used and not just resistors, is that the heat production of a resistive ballast would be huge, and you can't use the inductive kickback to ignite the lamp.

A similar thing - but in reverse because the load is capacitive - can happen with LED drivers with a capacitive dropper circuit. If you put enough LEDs with capacitive dropper on the circuit with the fluorescents with inductive ballast, you end up compensating the inductance with the capacitance and the power company is happy again.

So, I have an 18W fluorescent fixture with non-Philips preheat ballast that is labeled for 18W fluorescent lamp. But, when I tested it with watt meter, it says that it uses 39,7 watt. So my question is, does preheat ballast use more electricity than it's rated or because I'm using a non-Philips ballast that maybe it's for 40W tubes but labeled 18W and I'm over-ballasting it?

I think 18W is the rated lamp wattage and it doesn't include ballast losses. Also watt meters might be inaccurate when used with inductive or capacitive load.

The discussion about power factor and the 'capacitive' load of LEDs brings up a point. Many years ago, I recall it being speculated on here that the proliferation of CFLs and their less-than-high power factor would cause a noticeable strain on the grid. Has this been absolved in the age of LED?

some equipment have a watt and VA rating  the VA  is the "aperient" load and watts is the "actual" load
to understand power factor
AC power goes from zero volts up to the peak voltage and back down to zero in a smooth arc and a resistor across that would have the current it takes rise with the voltage and drop again in step with the voltage that is "unity" or 100 power factor
but a discharge lamp might for example start the discharge arc at 1/2 of the voltage rise and immediately start taking full power draw and at half the voltage will double the power drawn and that is what causes the "load" on the infrastructure

The discussion about power factor and the 'capacitive' load of LEDs brings up a point. Many years ago, I recall it being speculated on here that the proliferation of CFLs and their less-than-high power factor would cause a noticeable strain on the grid. Has this been absolved in the age of LED?

LEDs make the situation better, because they're such small consumers. For extremely small power consumption levels - think phone charger, LED lamp - there are no demands for power factor, at least not in the EU.
For bigger power consumers like PCs, you'll often find the words 'Active PFC' on the box of the power supply. One basic desktop PC draws as much power as 10 LED lamps of >1000 lumen. One gaming PC as much as 50 LED lamps. So therefore, it's much easier and more affordable to focus on those devices that really make an impact.

Nicer LED bulbs have a power factor correction circuit despite not needing one per law. There are driver chips that do it automatically. Big Clive has reviewed a couple of those.

In the past, capacitive loading of the grid has never been an issue. Inductive loads have been much more important to compensate. Our inductive street lighting ballasts are supposed to come with a compensation capacitor, and professional power users with giant motors also have to have giant banks of capacitors to compensate. So if individuals would use CFLs, it would only be beneficial because fewer of those expensive capacitor banks would be needed.
If there would be an excess of capacitive load on the grid today, the operator can just say 'Hey company, i see you have a big 500kW motor. Usually we demand you compensate that for 100%, but because of local grid conditions, we would like you to only compensate it for 50%'.