I wonder if these ballasts behave like HX ballasts or CWA ballasts.
Except their size and losses, they combine advantage of both:
Load characteristic close to the series choke/HX (so with increasing load voltage the current is sloping down, so suppressing the thermal instability of saturated vapor lamps), suppression of the mains variation (actually better than CWA use to; CWA is able to cover typical +/-10%, in extreme designs up to +/-15%, the mag-reg up to +/-30% feed supply variation).
Plus it could be easily made with an isolated secondary without that much of a cost or efficiency penalty.
But the "price to pay" is the larger mass and cost and higher ballast losses.
So because of that, their primary applications was highway HPS lighting, where long wiring made large voltage drops (the voltage stabilization suppresses that) and where the lamps are of saturated vapor type (HPS), even when the US market HPS are designed with the thermal feedback suppressed to some extend. Plus the isolated secondary allows the lamp shell to be kept grounded regardless how the input is connected (e.g. across phases,...)