VA and W are not the same thing - in AC systems, VA is a measure of apparent power, which is the vector sum of real power and reactive power (i.e. reactive elements - inductive/capacitive). W is a measure of real power only. The difference between the two is expressed as cos phi, the power factor, a ratio derived from the angle between the two values.

Circuit breakers and fuses are rated in Amperes, not watts. For a domestic supply, the main fuse is typically 100A which supplies a consumer unit which does have a main switch then usually the next overcurrent device will be the MCB or in some cases an RCBO which is specified for the individual circuit, not the installation as a whole which is protected by that fuse in the cutout, and maintained by the DNO.

If 9kVA were converted into current, on a 230V supply that would be just ~39A, maybe the sort of thing you'd see for a large circuit e.g. 40A MCB for a shower, but again not a whole installation.

The 'connection limit' or maximum demand, has always been expressed in either kVA or Amperes, because kVA is more relevant than kW when sizing conductors, protective devices, etc. It's nothing to do with being ripped off.

If you're out of phase by 90 degrees, yes you pass zero useful power and have a purely reactive load. Obviously that's not going to happen on a real circuit as there will still be a resistive element but that's how the maths works. The cosine of 90 = 0, hence the power factor is zero. At the opposite end of the scale, the cosine of zero (I and V in phase), you have a power factor of 1.

Edit: WRT power factor correction, there's more to it than just throwing capacitors at it, and it really depends on what the load is like in the first place. For a sinusoidal inductive load e.g. motors, capacitor banks can work well, but there are many other methods. On some loads like with switch mode power supplies you generally also end up with non-sinusoidal current as well as it having a reactive component - look up active power factor correction as an example of how it corrects the current waveform.