As you quoted, this is based on a completely desktop-statistical methodology relying on the assumptions that XMR distribution is a market phenomenon. For pre/insta/nonmineable coins this is not at all true. I believe it is true for XMR since I know its history.
Top 3 miningpools should be able to estimate number of wallets beeing in existance, seen Dwarf made a good job employing cookies to group miner's wallets, if they take network hashrate in relation to their own rate that gives you total number of miner's wallets.
Somehow this intersects with probably more precise numbers obtainable from inside Poloniex. Uncertainity on who is separating the mining from trading remains. But pure hodler who mine to not sell now (or tomorrow), are a very rare species. Their wallets would appear as beeing single additional entities.
At least there you get people with short time involvement added, who never bothered to even install the software. Unkommon behavier if you compare with cryptocoin involvement motivations, those count for probable new adopters.
Your completely desktop-statistical methodology relies on some inputs which are precisely known, like coins in existance. Number of wallets would add solidity to these data. I believe if one knows market dynamics, total supply, and number people involved this models reality with some decent accuracy.
