A loopstick antenna picks up the signal power from the field when the primary inductor has good resonance. That is, a reasonably high Q of the tuned circuit. In that scenario, the higher winding ratio would help, up to a point where the Q is largely limited by the quality of the inductor itself.
19 AWG wire would provide higher Q than thinner wires, provided that the ferrite material is capable of a high Q at the MW broadcast band. However, there are other possible effects, like how much of the ferrite bar is covered with the windings. Also, the Litz wire is also a way to achieve a higher Q at the high end of the MW broadcast band.
The winding ratio determines the impedance transformation. Your receiver input impedance is 200 ohm, so the "load" seen by the tuned circuit is 3200 ohm. That's high enough not to tax the tuned circuit's Q very much while achieving the impedance match (maximum power delivery) to the receiver.
Another factor is that the loopstick antenna picks up most signal power when the primary inductor is placed near the center of the ferrite bar. So, too many turns of windings may hurt beyond some point.
So, in short, the 4:1 winding ratio is a good place for a receiver impedance of around 200 ohm. Thick wire is advantageous, but a practical trade-off (size, weight, cost) is made with thinner wires in most AM radio construction. In particular, in superheterodyne receivers, the frontend tuned circuit is not the only element contributing to the selectivity (IFTs contribute much more), and the gain is also less important (it's easy to get more gain in the IF stages), so such a trade-off is very often justified. However, what you find is also reasonable when you try to maximize the antenna-frontend performance by itself.
