This insecure use of SIGHASH_SINGLE is useful for cheaply spending insecure low-value/dust utxos to miner fees – the idea is that you combine many inputs with only one output, use SIGHASH_SINGLE for inputs after the first so that you get cheap hashing, and for the signature R-value you use G/2 which has a short encoding, allowing you to clean up more dust in the same number of bytes. Because the utxos are already insecure, revealing the utxo’s private key by spending with a known R-value is fine, and because you don’t care about revealing the private key, having a signature that can be reused for other utxos with the same pubkey is also fine.
To some extent this can also be used as a “release” key; if you script is <Pn> CHECKSIGVERIFY <T> CHECKSIG then the T can publish a single signature that can be reused by P0, P1, etc who have locked different utxos against similar scripts (the Pn signature would be an ordinary SIGHASH_ALL here). That’s potentially more convenient than NONE|ANYONECANPAY as it doesn’t commit to the utxo being spent; provided you don’t mind having extra inputs so that there’s no output corresponding to this utxo’s input.
So personally I think it’s fine to leave this as-is, and just recommend people use p2wpkh, p2wsh or p2tr instead to avoid this weird behaviour unless they actually want it.