It looks like some good progress has been made
I had a quick look at the latest wav file and it shows that the 50Hz modulation level is much lower. However, there is now some 50Hz baseband 'hum' on the signal as you already spotted and this looks like it has a square wave spectrum as it looks like 50 Hz with a lot of odd order harmonics.
The PLL loop response is also different with a wider loop bandwidth and I guess this is due to your retrofit of the original loop filter parts. However, RX2 doesn't have this wider loop BW? There now looks to be some noise peaking in the RX1 loop response and the loop BW is beginning to look a bit wide with less damping. It's beginning to look quite wide wrt the 1kHz phase detector frequency and if it gets much wider or shows less damping (as N changes across the tuning range of the VCO) then it could mean the loop stability could become marginal in parts of the tuning range of the Digital VFO. The 1kHz reference sidebands also look to be lower now. However, the wider loop BW does mean the loop is now able to cancel/correct the 50Hz pickup as an error signal. The 150Hz terms are now the highest and this is because the loop can't correct this higher frequency out as an error signal. So I think a tiny part of the improvement is also due to the loop filter changes.
In terms of the 50Hz warble, it does look and sound a lot better and it would be interesting to see if the transformer arrangement in the second receiver can be improved if you get the opportunity to look at it.