More than the psychological factors which you talked about (i.e. that people are weary or governments have declared triumph over the virus - which is false concerning France for instance), I think that the main reason is the seasonality of the disease (which is a well-known phenomenon observed for other diseases, as far as I know).
I think one can see it observing Sweden's case: the number of infections (and deaths) had slowly diminished starting from may (not as fast as in other European countries after the lockdown) and the deaths have been rare (1 or 2 per day) until now. But now as everywhere else infections are increasing.
Furthermore, it is not really credible that most European nations have had coincidentally a resurgence in cases at the same time. As far as I know, it is often quite complicated right now to move from a country to another at least in my case (France), so I don't think this can be explained by spreaders from a country to another (and there were still clusters at that time (early September), so why would transnational spreaders be necessary to sparks infection anew ?).
I came back thinking again about this and I really struggle to see how it can be a great trade.
Your theory as far as I understood it is that this is a sort of one year option betting on liquidity constraint of the traders on polymarket.
I say option in the sense that you are burning through something to hold a right. In option theory that's called theta : the derivative of the option value to time. If you are long the option your theta is negative and you are losing money just by holding the right. In this case the holder of the yes is short theta (yeah the term is unlogicaly overloaded) in the sense that as the year progresses the annualised earning from the holder of no is better and better. If you can buy a no in december for .97 you're making a great trade because you would be earning (100/97)% for a month which is 44% annualized.
So on a first order rationality (just considering it as a loan) the value would be decreasing until then with the equation linking the value of the yes so that it equates the yield you can lock with a same expiration TBill. (It should rationaly be more - as people have pointed out - because you hold a more important counterparty risk and have liquidity problems).
As of now a 1 year TBill is about 4.0% annualized which is really more than what a current 97.3 yes for 9 months holding would pay you (3.6%).
So as of now you might say : ah the holder of the no should have better opportunities (tbills are more liquid and have better liquidity) eleswhere so they should be selling their no (thus rising the yes) and buying tbills instead. Realistically obviously a change of .4% annualized on 10k is 40$ so not worth the hassle (and you could actually not get the full 10k out at the same yield because of liquidity).
This is the baseline I would take to think about the time value of money here. And according to this model the time value of money would be directly influenced by the interest rates of treasuries with another effect accounting for the people having put some money here to wait.
The first effect if you wanted to trade it you should rationnaly just buy (leveraged) yield spreads and stuff like that.
If you want to bet on the other effect I struggle to see what is your edge ? It was probably not a good rational choice to put money into this even at 4% earlier this year. So you are betting that their behaviour is gonna be ah now I can trade other stuffs let me remove these no. But what is your expected payoff from this effect and what makes you confident this payoff is positive ?
In summary of this trade :
Would be happy to discuss as it is very possible (20/30%) I have made an understanding mistake somewhere as my knowledge of prediction market and their payoff is very weak.