The key detail of Bitcoin's use of blockchain is to enable mutually distrusting groups to agree on some state in the absence of any centralized authority. In the above case, as in so many buzzy uses, the use of some blockchain to transact seems redundant since the value of tokens being exchanged are entirely dependent on a central authority, and could much more easily be facilitated by that authority (this is just the common "99% of blockchain uses could just be a database" argument I guess).
Another aspect is that when the central authority is also the sole issuer of tokens, no counteracting forces exist to drive a meaningful market dynamic. No matter how much price pressure exists from users, the authority can always act to restrict issuance or increase issuance to force a particular price, and over the long term in the absence of that authority (due to bankruptcy, obsolescence, ..) the value of all issued tokens drifts towards zero.
On saying that though, there seems to be some promise one step removed -- tokens that activate functionality in software produced across a wide range of vendors. In the case of a game, that could perhaps be realized as something like a common set of power ups tied to NFTs that could be reused in multiple games. The question is how you'd incentivize multiple vendors to buy into the same set of tokens, given doing so is likely to have some real world cost attached to it, and obviously introduces a creative limitation to game design.
Meanwhile, can't help but agree, the current NFT applications (million dollar web site clones, weird eBay for JPEGs clones, ..) seem entirely pointless and more a symptom of something than a solution to anything.