I think the terminology may be the problem here. You were actually measuring idle current at the cathode (90mV) rather than bias voltage at the grids, correct?
In that case, it was 'idle current drift' rather than 'bias drift'. Bias drift will always cause change in idle current, but idle current drift is not always caused by bias changes. Hope that makes sense.
Change in plate or screen voltage can cause change in idle current, this cap could have caused that. Hum/noise could also be seen as change in idle current by the meter, and this cap may have caused that too.