Nope, I think it does not. The 100nF of course forms a low pass filter with any resistor in line but it won't be able to source a static current in the range of some 100µAs. A current source is the only explanation for a voltage at the RST input which is different from the calculated one based on the 10k pullup and 1k5 switched to GND by GPIO16.
@davydnorris & @schufti
Yes, fully agreed on the diode recommendation. I'll give that solution a try in the upcoming days. Connecting the anode to RST and the cathode to GPIO16 really looks best.
Indeed I've not thought about that serial reprogramming issue as my device was already in a state where I used OTA updates all the time and I was focused on getting rid of this instable deep-sleep behavior.
Until today where I bricked my OTA loop, and oops...