Well, improving the power factor in the 50/60Hz domain would require storing considerable amount of energy (due to the loooooooong period) in the inductor. So, you are throwing away the BOM cost and size benefit.
For a real solution, there is no way around of using an SMPS. This way you can minimize the solution size and cost, by:
- utilizing the DC link capacitor better (by allowing ripple in the rectified voltage, and still drive the LEDs with constant current, eliminating flicker without massive capacitance)
- minimize the inductor size, by using high frequency, so it needs to store much less energy, yet allowing active power factor correction.
Granted, this is more complex, and there are more chances of getting it wrong. OTOH, it's not a rocket science, it's a solved problem, and most crappy products are crappy on purpose.
Trying to make the cheap, simple, crappy capacitive dropper better by adding expensive parts makes very little sense. Use it, in its simplest form, for its simplicity and low cost whenever applicable.