Designing around say 10 or 15A RMS seems reasonable, and having a saturation current more than twice that means you don't have to worry about a crest factor of the same proportion (i.e. 2:1)*. More would be nice; that can be done by using more core and gap (= higher saturation current), or less wire (or using the same wire but just rating it more conservatively, in either case giving a lower maximum RMS current rating).
It doesn't seem very useful to just go and waste (RMS) current capacity. So I think I would use 10-15A wire, and a 30A+ saturation rating (at inductance -30% or better), and just make sure to check that peak current isn't going into the concern range. Maybe get a mains frequency current transformer, or Hall effect probe, and verify that load current is within these ratings, when in use.
*I forget if crest factor is defined in comparison to square wave (RMS = average = peak), or sine (peak = sqrt(2) RMS). If the latter, then that'd be a ~1.4 crest factor, whatever.
The 250uH choke, I mean, it's literally five times bigger than the 50, so if you want the same rating, wire and fill factor, you're going to have to use a stack of five cores. You can vary the gap and wire, but it's probably better to keep the gap nominal, whatever nominal is driven by.
(Which, because your winding factor is a bit low, what with all that insulation there -- a somewhat higher mu_e may be desirable, maybe 30 is ok, maybe 45-60 is better, give it a try and see what seems to fit best.)
Regarding winding area, keep in mind that you'll be lucky to fill more than half the area with wire -- that includes area wasted to the gaps between round wires, tape, bobbin, etc. A bit better if you're not using a bobbin. And that's just enameled wire -- a winding factor more like 0.3 might be reasonable with jacketed wire. In any case, do some figures with turns/layer and number of layers, see if you can pack it in there with a little free space so it's not totally crammed in there. That's your maximum turns count, and if you're heart set on using 3 and 1 core stacks for the two chokes, expect to fully load the triple stack, and only 60% load the other.
Which means, sure, you could fit more turns on the 50uH's core, it's got 40% wasted space; but so what? If you increase turns, you increase inductance and decrease Isat; if you increase gap to compensate, L is still higher (for same Isat). Why bother, y'know?
![Smiley :)](https://www.eevblog.com/forum/Smileys/default/smiley.gif)
Tim