Hello Bertil,
I might have expected the lower external capacitance value to be connected to the BDM pin, and the higher value to the reset pin. Keep in mind that the pair of schottky diodes are likely to provide about 20pF shunt capacitance. I don't think you need any more.
For the reset pin, I would typically use an external pullup of 10k, and a shunt capacitance of 10nF. This gives a time constant of 100us. I shouldn't expect any POR issues with this delay. Even using an internal pullup (say 30k-60k), the time constant would remain less than one millisecond. What is the oscillator stabilization delay for the device you are using?
I am not sure why you have concern that the reverse breakdown voltage for the schottky diodes may be too high - if this is what you mean?
There may also be some misunderstanding about what you are trying to protect against. I don't think there will be any issue while the BDM lead is connected, since externally applied impedances will be relatively low. The more susceptible situation would be when the BDM lead is disconnected, with the possibility that external electrostatic impulses may reach the pin(s). However, I might also assume that the BDM header would not normally be exposed to the "outside world" during operation of the equipment, only during factory programming and test. Other pins on the MCU, such as ADC inputs may be more susceptible to external interference, and may need more stringent protection measures.
Regards,
Mac