Nobody asked you, Erik, but, of course, you had to spout on anyway.
which begs the question "since nobody asked you, why did you sprout on?"

phone calls that told you that your circuit had to be power-cycled twice in order to get it to start it seemed to work because the phone calls stopped coming.
1) we have tens of thousands of these out there
2) the phone calls were not "random" but mainly from our most demanding customers.
3) the typical customer will have between 10 and 500 units, our most demanding customer has several hundereds.
4) It was not necessary to call these customers, from other expreience, I know they would have called again if the problem had not disappeared
5) thus the statistical probabilty of the solution being correct is aboiut 99.99%. Of course, you are going to come back and harp on the percentage, go ahead, I'll ignore that.

Then, when asked how the problem was addressed, your response amounted to, "well, my original circuit didn't work right, so I designed another one
Adding a proper supervisor to $100 worth of other components is hardly "designed another one". Had I made other changes at the same time, I could not argue that adding the supervisor was the solution.

Your only observations were those phone calls, and your solution was to slap together a different circuit
reasoning about what heppens at power on and drawing a conclusion, then implementing it, is hardly "slap together a different circuit".

I appreciate that doing anything that I do not do Richards way makes me a bungling amateur, so be it.

Once again everybody, except you have realized that a proper supervisor is mandatory for flash based circuits.

Erik