This reply is more speculative. I'm not a geologist, hydrologist, or have any particular knowledge about fluid dynamics.
Your suggestion that the water could concentrate radioactive contamination is interesting, but it doesn't feel right to me. Let me propose a thought experiment.
Let's assume 1 kg of soil naturally contains 370 Bq of K-40. 1 Bq = 1 disintegration per second.
First, K-40 is also a gamma emitter. 11% of the time, it decays with a fairly high energy gamma. So we would have to assume the lab failed to notice the gammas. For the sake of argument, let's make that assumption.
A 75 cm^2 sample of Dubinina's brown sweater measured 9900 disintegrations / minute / 150 cm^2. So that sample alone contains 9900 * (75/150) / 60 = 83 Bq of radiation.
At this point, one might conclude that the sample was only contaminated with the radiation that 223 g of soil naturally contains. So if 223 g of soil wash over that portion of her sweater and deposit all their K-40, the mystery is solved.
But that isn't really realistic. The 223g of soil contains a trace amount of K-40 and, well, almost 223g of soil. I see no reason for the K-40 to stick to the sweater but the rest of the soil to pass on by.
If we instead assume that K-40 and soil both deposit on the sweater in equal proportions, then that 75 cm^2 sample would need 223g of soil on the sweater as well. To put that in perspective, that much dirt has to be on an 8x8 cm piece of fabric. Or for us Americans, a half pound of dirt on a 3 1/2" x 3 1/2" piece of fabric. That's substantially more dirt than sweater!
Also, you said "Indeed i would argue that the results are inconclusive without also measuring the silt underneath the bodies in order to properly determine the background level."
There is something else interesting in the report. Table 2 item 1 lists "Soil from No 1" (Kolevatov). I am guessing that some soil came along with the clothing sample and may have been brushed off and placed in the detector. The radiation measurement was 96 counts per minute. Background for the instrument, measured at two different times, was 90 and 100. So this soil is not significantly higher than background radiation.
I caution against reading too much into that. We don't know where the sample came from, or how much soil was there. (Does the original Russian document have any clues?) Also, beta radiation is easily absorbed, so a clump of dirt can "self-shield", lowering the observed beta reading somewhat. But still, this soil clearly was not screaming hot.
My gut feeling is that I can't see any natural explanation for the radioactive contamination on the clothing. The examiner concludes the same in the interview. "Should there be (can it be) increased contamination of clothing with radioactive substances under normal conditions, i.e. without being in a radioactively contaminated environment or place? Answer: Definitely not."
I don't believe water will concentrate natural radiation. If anything, I think it will dilute contamination.
This doesn't mean the clothing must have been contaminated pre-mortem. I can see post-mortem options with contamination transported by water. But in that case, my gut feeling is that there would need to be a whole lot of artificial contamination dumped on that area, as only a small percent would deposit on the clothing and stay there, e.g. not get washed off.
Still, I think it would be easier to assume the clothing was contaminated at, before, or shortly after death than during the three months afterwards.