After some more testing, the issue seems to be related to the visual gain setting at first start. After setting the visual gain to 0dB (and slightly adjusting the low dispaly levels), I do no longer see the issue when manually changing the RF-Gain or when stopping an restarting the radio. Interestingly, after changing the visual gain setting from the default setting of -30dB, the "Radio PlutoSDR> Error writing Channel attribute hardwaregain = 50, error 95, Unknown error" does no longer show up in the logfile. The other error is still there but seems to be related to some other issue.

@Sigi; I'm fine with 50dB hardware gain, as I can lower the gain to 30dB before the SNR starts to decrease. So there is plenty of headroom with 50dB hardware gain (at least with my LNB setup and some 20dB of additional attenuation in the LNB interface).
As far as I know, signal levels of 54 dB over noise are technically not possible on the qo-100. Adjusting the gain for a noisefloor at 0dBuV and then switching the scale to S-Units shows indeed around S9 for the beacon - however  the noise is at S3-S4 with that setting. I do not see much value in using S-Units on the qo-100 as you can fiddle with the gains to get whatever value you want. The IARU recommendation for S9 differs between HF and VHF, setting the noisefloor to -20dBuV results in a more realistic S5-S6 for the CW-beacon in respect of the beacon SNR.

@jdow: I hope you are wrong when you suspect that new features introduce new issues in unrelated parts of the SDR console. For me, this would be a clear indication to step back and seriously question my development process. 

73 Jean DJ0VL