Well, 50% drop of the panel with 3 diodes is hardly possible. It's either 33.3, or 66.6 or 99.9% drop. But let's not be nitpicking.
Now, if 33.3% of one panel is shaded... Voltage of this panel drops by 33%, voltage of series 3 panels drops by 11.1%. Amps of the shaded panel don't change, yes.
Amps of the string of 3... Don't change? I remember discussing this with the maker of my MPPT controller - back in those ancient times 4 years ago when the US-made electronics still existed. His point was that it depends A LOT what part of the panel is shaded. If this is the end of the panel, i.e. all 3 sub-strings are partially shaded, this creates a bottleneck so bad that MPPT has little to grab on. For the entire panel, and thus for the entire string. Output drops to almost nothing.
Back to the perfect scenario with exactly one sub-string shaded, i.e. 33.3% V drop on this panel, 11.1% V drop of 3 panels. Is there such thing as 33.3% R increase of one shaded panel in the string of 3?
PS: Upon running some numbers - 2*24 panels with 500W total could still be wired in parallel using off the shelf MC4 extension cable #8 (yes, there is MC4 #8). On a small trailer/camper the V drop will be under 1.5% on 20ft run. No junction box.