Unless you can present more reliable numbers, 100K D3 is probably close enough.
There is no magic that turns a bad estimate into a good one, just because it's the only estimate available. If we're standing on a bridge, and I tell you that, in my opinion as a random dude with no engineering or medical training, there's a 90% chance you'll survive if you jump off, and you say, "I'd actually like a more knowledgeable estimate," how do you feel when I say, "well, since you only have my estimate, it's probably close enough."
Is my estimate better because it's the only one? I can't quite understand why this seems hard for people to understand. Bad data is bad data. The presence or absence of other data doesn't affect that in any way.
Initially Nikon was producing 8K D3 a month and due to high demand, they increased to 9K a month. Those numbers came from Nikon themselves. I would assume the D3 sales was skewed towards the early part of the production cycle and it had to slow down after 6 months to a year.
You assume. By how much did it slow down? Did it slow down after 6 months, or after a year? The difference matters.
The numbers you provided actually give hints that the serial number database may be off by a wide margin on D3 sales. Nikon's initial estimate of sustained sell-through was about 100,000 units per year (8K per month). After 3 months, they increased that estimate to about 120,000 per year (they upped production (a considerable investment) to 10K per month, not 9K). The camera was in production for over 2 years.
If the 100,000 figure derived from the serial numbers is correct, that means Nikon was off by a factor of about 2.5X in its production plan. Possible? Yes. But there's no evidence that was true in the case of the D3.
To me (and I worked in the industry for 15 years and have direct experience with flagship camera sales curves) the better guess is that Nikon probably sold somewhere in the range of 150,000 to 200,000 D3 cameras. That is an educated guess, but I would never claim it's anything better than that. And I would never use it to make arguments about lessons Nikon has learned or about how much of any camera Nikon should produce in the future. It's way too unreliable for that.
If my guess based on Nikon's production figures is right, then the serial number database is off by a factor of 1.5-2X. That's very poor precision.
For the D3, Nikon only had one sequence of serial numbers for their world-wide distribution.
This is also an unverified assumption.