Hello Dave! It's Jacob from the comments section of the video. Thanks for posing this, it looks exactly as I thought it would. This is perhaps worthy of another video. The harmonic content here is horrendous, this is what is causing the poor PF. This is also the worst type of poor power factor not only because it's difficult to correct, but because it causes interference and a host of other noise-related issues. As I stated in the comments section it can be corrected with filters in the power system but these are usually not found until a transmission substation after all the losses have occurred.
Here is a quick wikipedia page describing the components which will ultimately end up removing the harmonic content. The line reactor section describes the use as harmonic filters.
https://en.wikipedia.org/wiki/Current_limiting_reactorWhy does harmonic content create low PF??
Most people are familiar with P = Vrms*Irms*cos(theta) where theta is the angle between the voltage and current waveform. This equation holds true HOWEVER it can only be applied (without thought) to pure sinusoids of the same frequency. That means the voltage and current waveforms are of the same frequency and are both sinusoidal. For any non-sinusoidal waveforms such as this current waveform, they must be decomposed into sinusoids using the Fourier series and analyzed using the same equation for every harmonic. I will post an image of this formula from one of my power electronics textbooks.
One important realization is if the voltage source is a sinusoid but the current is not, ONLY the fundamental frequency of the current waveform can draw real power. The harmonics are orthogonal to the voltage fundamental frequency and will never draw real power unless they have a voltage harmonic to pair with. These harmonics do still draw current and contribute to the overall RMS current draw. These harmonics of different frequencies are responsible for the apparent power draw and the power factor is calculated using its fundamental definition PF=P/S. This is how PF must always be calculated for non-sinusoidal waveforms. P = Vrms*Irms*cos(theta) WILL NOT work in this ball field. Therefore adding inductance or capacitance to compensate for the PF will not work, the harmonic content must be filtered out.
This one of the reasons is why there is generally so much regulation on harmonic content. If these power supplies were a little larger (say for a TV or PC) they likely would not be permissible by many regulating entities with harmonic content of this magnitude.