Simulations using EZNEC V6 confirm that the azimuth radiation pattern for a quarter wave vertical with 2 radials is little affected (0.5 dB max) by the ground going from perfect conductor to very lossy. The deviation from a perfect omnidirectional pattern is less than 0.5 dB for 2 radials cf 4 radials.
Going to three or four radials is fine philosophically but in practice on a summit is a time waster and hard to do in well vegetated summits. I have often operated on the side of a track and having a radial crossing the track is a concern.
The elevation pattern changes of course as losses rise with the low angle radiation falling away as the ground losses rise. Lifting the radials reduces the losses a bit but does not restore the pattern. (3 m max assumed, ground still lossy).
DX signals (16 degree elevation) are 6 dB down for lossy ground typically. ERP is down about 6 dB fof a lossy ground compared to a lossless one.
Interestingly for a 1 radial quarter wave radiator the best compromise seems to be to run the radial along the ground. The front to back is about 6 dB but raising it to 9 feet agl increases the F/B ratio to 11 dB and increases the amount of signal going vertically up. Comparing the radial on the ground with radial 9 ft up, the signal at 16 degrees varies by about 1 dB. For a radial on the ground/grass the set-up time is reduced and no need to find supports or have long cord extensions in order to keep it above head height. A shorter mast can be used.
Should you be struggling to work a station in the null direction I’m sure you could grab a radial with an alligator clip and have it deployed in 45 seconds and then have a 6 dB or more lift.
The length of the radials for resonance varies by almost 2 m when different elevations and different ground conditions apply.
Therefore having an ATU is recommended due to the wide range of ground conditions likely to be encountered especially in the SE VK Alps.
Adding a loading inductor shortens the antenna and can be used on the radial(s) as well of course but bandwidth shrinks and losses are higher. Whether it is worthwhile is a matter of individual choice. getting a wire 17 ft up on any summit should be possible.
For the same lossy ground my 6.7 m/side doublet (7 m at the centre inv vee) is comparable to the vertical broadside on and about 6 dB down off the ends so compares surprisingly well with the one radial vertical with the advantage of being multiband.
It is a bit harder to erect in the forest. I could of course run one leg vertically and the other just on the bushes if it was a tricky QRH. The simulation shows this is not a bad option for 20 m with just the one null off the back, away from the horizontal wire as expected. It is viable for stations no closer than 150 km on 40 m.
I think my work is done now.