Yes, but only proportionally to its impact on load regulation as well, and it has to withstand nearly 120VAC at whatever current it passes, which means a choke as big as the transformer most likely.
To put it another way: you're trying to dampen a current surge, but the desired current flow is itself alternating. If you attenuate a big peak one way, you can't avoid attenuating the smaller peaks in either polarity. Not with a linear element.
It almost works on DC, since DC is DC, and transients are AC. Trouble is, since the load is capacitive, attempting to use an inductor will only increase overshoot. It needs to be heavily damped, meaning you need ESR in the loop. With low inductance and large, lossy capacitors, the ordinary short-circuit event is destructive (100s of amps) but otherwise fairly well behaved (little overshoot), because ESR > sqrt(L/C).
Say for a typical example, there's a 1m twisted lead from automotive power / a 12V car battery, to a modest power amplifier that draws a few amperes. The supply is maybe 1uH, and 1000uF would be reasonable bypass. Such a capacitor might have ESR ~ 0.05 ohm, but sqrt(1u/1000u) ~= 0.03 ohm, so the ESR is dominant and it will go "thud" when connected. It'll also draw a peak of about 12V / 0.05 ohm = 240A, which will result in a snappy spark and that's all.
If you use a lot of inductance, you do reduce the peak current. Suppose we put a "hash filter" on here that's 100uH. Now sqrt(101u/1000u) ~= 0.3 ohms and it will ring and overshoot. But the peak current has dropped to about 12V / sqrt(L/C) = 50A, so it is true that inrush will be improved. It's just not a real attractive way to do it. Oh, notice that also assumes the hash choke is itself still linear at 50A -- unlikely given it only needs to handle a few amps DC. It most likely won't be linear, and therefore the same exact problem that affects the OP's toroidial transformer will pop up here: spiky inrush due to saturation.
Tim