Doppler correction is not good and M6P frequency drift
The doppler correction works fine. The issue is that M6P wasn’t transmitting at that time.
What you see on those two observation and some others at the same time is 43793 CSIM-FD satellite. For example at https://network.satnogs.org/observations/1645146/ the red line shows what CSIM-FD signal was expected to follow, as you can see it matches perfectly the signal you see:
We need to be careful on vetting for M6P as CSIM-FD follows it these days.
For example here the signal seems to be in straight line https://network.satnogs.org/observations/1639140/ but it is CSIM-FD, you can also say from how the signal looks on the waterfall and that it starts near the end.
For comparison check this good M6P observation: