We've had pruning since ~2013.
Yes, but you cannot do Initial Blockchain Download from a pruned node, at least not yet. Which means, that we still need that data anyway, no matter if they are behind OP_RETURN or not.
And my message noted as much: Someone needs to store it. But this isn't an issue: Many people are willing to do so and no one is forced to do so.
To be pedantic one can also bring a node up without accessing that data with assumetxo (or a utxo snapshot), making the chain before that "purely virtual" to use your terminology. I didn't bother mentioning it, because I think your comment about people being forced to store things is amply addressed by fact that only some people need to store the whole history and many do, so you don't have to store it if you don't want to. It's a choice.
If you disagree, then why draw a line at "embedded data" and not the voluminous junk that isn't embedded data-- like the gigabyte after gigabyte of satoshi dice 'you lose' transactions? At the end of the day anyone choosing to store all of Bitcoin's history is choosing to store an absolute mountain of junk that is of no interest or relevance to themselves, that's just the nature of the beast. We all set the terms to confine the rate it can increase in order to protect the viability of the network.
In theory, such things could be achieved through commitments, but they are not widely adopted,
Not so, commitments are extremely well understood now -- and OpenTimestamps alone does more timestamps a day than Bitcoin's entire transaction volume.
Particularly given the considerable transaction fees on Bitcoin (even at minimum fee rates) it's clear that anyone embedding data specifically wants it to be in bitcoin. The argument you were advancing only really made sense more than a decade ago when fee levels were negligible-- that's fighting yesterday's war. Congrats: it was won. No one stores any kind of data in bitcoin because it's functionally free to do so anymore, when they store data in bitcoin it's at great expense and for reasons the transacting parties feel justify that expense.
In the context of the NFT crap that gets the most attention they *want* to pay high fees to make their issuance scarce. In the context of ZKP additions they need guaranteed publication atomic with some transaction. If what you want can be done with a commitment you use OTS and pay absolutely nothing (or some random altcoin and pay virtually nothing).
