I don't know how formed impulses in MN310x. but I can tell, how to obtain the desired wave shape in a makeshift device.

If the MOSFET inverter will square wave, the output will have the same momentum, but reverse polarity, which edge will nearly match the fronts of the input pulse.

But if the inverter put integrating RC-chain, the fronts of the input pulses are integrated, and lose a rectangular form. The inverter is switched with a little delay, the amount of which depends on the time constant RC-chain. Output impulse will be detained for a short time, but its duration will remain almost unchanged. In the figure shown in the top three charts.

We need to slightly reduce the pulse duration. To do this, add parallel R diode. Now rising edge input impulse will be integrated, and falling edge remains rectangular. On this inverter switches with delay only on the rising edge. Pulse duration decreases latency on the rising edge. The figure is the fourth schedule.

Now again invert this momentum, and we get what we need for CP1! Of course, the last inverter need to collect two parallel inverters from the 4049.

The same operation will do for the momentum of the second phase of the management of BBD.

We need to reduce the pulse duration at time t = 100 ns.

We will consider the capacity of RC-circuit equal 20 pF. This is the input capacity of each inverter 4049. This means from external capacity, you can choose.

Then the resistance RC-circuit is: R = t/C = 10^-7/(2 * 10^-11) = 5000 Ohms. The nearest size – 5,1 kOhms.

If necessary, time can be adjusted by modifying the Tdead value R from 1 kOhms to 10 kOhms. In my view, the optimal delay is ~ 50 ns. With R = 2,4 -2,7 kOhms.