Take a look at the datasheet again, in the one I found there is a jumper J1 on the module that would prevent a "reset" with the pin. It is not a real reset input, it does a complete Off/On cycle for the chip, switched with a P-channel FET, which is shortable by J1. 100ms might be too short, it depends on the used capacitors and other factors.
I would start with a few seconds reset-pulse, decreasing till it begins to fail the reset and then double that time.
Edit:
There might be another reason for a non working reset: The "Select"-pin might power the module if it is high during a reset-pulse. In the schematic of the DK it is unconnected, so this should not interfere with a proper reset. I would check if the pin is really unconnected in your revision.
But any other interface-pin that is directly connected to the BT-chip on the module would probably have a similar problem, ensure that all this pins are actively driven low shortly after setting "Reset" to high, stay there during the reset pulse, just before setting "Reset" to low again the inputs on the controller-side should be switched back to inputs.