I checked the data again:
It seems that a few minutes after the BMS-current exceeds maxChargeCurrent, the MPPTs left the MPP of around 100V. The voltage then rose to higher values. Maybe they reduced their current but because of an unknown reason, the Victron system regulated the current with a constant offset on maxChargeCurrent of about 2 to 3Amps.
Here the raw data from influxDB without aggregation:
So the charge current limit seems to work in general, but with an offset.
For further information:
The maxChargeCurrent is unstable in this condition, because another self programmed regulator trys to limit the max cell voltage at cold conditions to 3.4V, which worked in general:
It really seems, that there is only an offset problem. Strange!