Most (or all?) of the North American 2x40 preheat ballasts had one tube delayed by a cap, or, as some manufacturers referred to it, "lead-lag" or "stroboscopic corrected".
That's interesting! I think that magnetically ballasted fixtures here were manufactured just with power factor compensation capacitors and suppression capacitors in starters (AFAIK).
Wikipedia Fluorescent lamp page shows the "beat effect" video (it's almost at the bottom of this page) and both tubes seem to have the same beat - so no stroboscopic correction.
In Europe, just the series inductor ballasts are nearly everywhere, mainly because they are very simple. So there the high power factorpower factor is indeed just the capacitor parallel to the mains.
In the US, way more variety of ballasts are in use, as the 120V does not allow that simple series choke for a longer lamp.
Most ballasts were of the RS type, because the 120V needs a transformer anyway, so adding the extra filament windings was cheaper than the starter and it's accessories.
Cheap ballasts for domestic use (NPF) just the high leakage autotreansformers, just because there was nothing else necessary. These ballasts were lag only and usually supplied the lamp by way lower than rated wattage (at home, such high output as from 40W lamp was not necessary, but it was a benefit to use that most common lamp type, because it was cheap). Because of the main use, there was no need to compensate the power factor there. Because mainly single lamp, it had the 120Hz flicker.
The commercial use units used full lamp power (because the light output was really needed) and it was arranged so, the power factor was corrected before reaching the primary winding. SO it was either lead style ballast with an air gap in the transformer core (so it's low main inductance then compensated the power factor of the capacitive load), mainly used with ballasts either for single lamp or two in series.
This was lead only, so no 120Hz flicker reduction (just the end 60Hz flicker in the electrode area was suppressed by having the current flowing in each tube in the opposite directions)
And then some were lead-lag. Mostly the four lamp units (two series pairs), and some two lamp ballasts. It was a bit more complex core than for the plain series combination (the lead-lag requires three sections separated by the magnetic shunts: Primary and two secondaries), allows independent operation of each branch and partially suppresses the 120Hz flicker.
So if the ballast is designed to be high power factor on it's own, so no separate power factor correction.
And then of course the high frequency ballasts took over on both sides of the pond...