We can't know the actual number but we can come up with a reasonable estimate. The chainwork value is a good estimate and a standard deviation value tells us how good that estimate is.
The chainwork estimate and standard deviation values account for those possibilities.
The next step would have to be what a 'reasonable estimate' is for the hashpower that is out there that we don't know about.
I really think it's a lot more then people think OR can be proven. But, I [wait for it] can't prove it.
We know about the big pools, we know about the small pools, we know about the occasional unknown block that pops up.
Most people think that the number of people solo mining to pools they setup themselves is small as in dozens at most. I think it's way higher then that. But we never see them because the "paranoid tinfoil hat brigade" that does not trust mining pools and thinks they can do better themselves because all the pools are out to get you are a lot more common then we see here.
But, we will never see them get a block or even know they exist because their crap setups could never find a block even with all the hash in the world. But they are doing it themselves because they know better. You see these people pop up in the hardware and pool board now and then.
In terms of BTC itself they really do not matter, but and I guess we will have to disagree here, saying that you can use stats to find them is not a realistic thing because they are hashing to find a block but they never will and cannot and there is no way to even know they exist.
I can take you to a data center in NY where someone is paying probably close to $1k a month to run a bunch of S19 that mine to his super special custom coded pool that is going to get a block real soon now.......
-Dave