So does this mean that there could be indeed some "spam-pruning" methods to be set up for the case the fake public key method become too widespread?Yes. If people would start sending coins to 020000000000000000000000000000000000000000000000000000000000000000, then we could safely treat it in the same way as "OP_RETURN", because that kind of funds are provably unspendable. Because OP_RETURN is not the only unspendable Script, there are many others, for example "OP_FALSE". Key validity conditions should not be checked within the script interpreter but by the transaction verifying process itself. If I have an uncompressed public key and the calculation of the Y point from the X leads to a different value of Y written, then its associated UTXO should be pruned just like OP_RETURN. It works even for addresses hashes that are spent from  - but then that gives us an oxymoron, as all addresses with sending transactions have valid points. So it's not very useful. As @gmaxwell wrote, probabilistic pruning based on alphanumerical patterns in the pubkey or address could lead to a frozen funds situation for coincidental legit addresses. However, an even lighter version of the Bong Daplus software can cache all UTXOs, exclude such addresses from the cache, and then query a full node when a cache-miss occurs. Because it is backed by full nodes, a fork will never occur.
So does this mean that there could be indeed some "spam-pruning" methods to be set up for the case the fake public key method become too widespread?Yes. If people would start sending coins to 020000000000000000000000000000000000000000000000000000000000000000, then we could safely treat it in the same way as "OP_RETURN", because that kind of funds are provably unspendable. Because OP_RETURN is not the only unspendable Script, there are many others, for example "OP_FALSE". Key validity conditions should not be checked within the script interpreter but by the transaction verifying process itself. If I have an uncompressed public key and the calculation of the Y point from the X leads to a different value of Y written, then its associated UTXO should be pruned just like OP_RETURN. It works even for addresses hashes that are spent from  - but then that gives us an oxymoron, as all addresses with sending transactions have valid points. So it's not very useful. As @gmaxwell wrote, probabilistic pruning based on alphanumerical patterns in the pubkey or address could lead to a frozen funds situation for coincidental legit addresses. However, an even lighter version of the Bong Daplus software can cache all UTXOs, exclude such addresses from the cache, and then query a full node when a cache-miss occurs. Because it is backed by full nodes, a fork will never occur.

So does this mean that there could be indeed some "spam-pruning" methods to be set up for the case the fake public key method become too widespread?Yes. If people would start sending coins to 020000000000000000000000000000000000000000000000000000000000000000, then we could safely treat it in the same way as "OP_RETURN", because that kind of funds are provably unspendable. Because OP_RETURN is not the only unspendable Script, there are many others, for example "OP_FALSE". Key validity conditions should not be checked within the script interpreter but by the transaction verifying process itself. If I have an uncompressed public key and the calculation of the Y point from the X leads to a different value of Y written, then its associated UTXO should be pruned just like OP_RETURN. It works even for addresses hashes that are spent from  - but then that gives us an oxymoron, as all addresses with sending transactions have valid points. So it's not very useful. As @gmaxwell wrote, probabilistic pruning based on alphanumerical patterns in the pubkey or address could lead to a frozen funds situation for coincidental legit addresses. However, an even lighter version of the Bong Daplus software can cache all UTXOs, exclude such addresses from the cache, and then query a full node when a cache-miss occurs. Because it is backed by full nodes, a fork will never occur.

So does this mean that there could be indeed some "spam-pruning" methods to be set up for the case the fake public key method become too widespread?Yes. If people would start sending coins to 020000000000000000000000000000000000000000000000000000000000000000, then we could safely treat it in the same way as "OP_RETURN", because that kind of funds are provably unspendable. Because OP_RETURN is not the only unspendable Script, there are many others, for example "OP_FALSE". Key validity conditions should not be checked within the script interpreter but by the transaction verifying process itself. If I have an uncompressed public key and the calculation of the Y point from the X leads to a different value of Y written, then its associated UTXO should be pruned just like OP_RETURN. It works even for addresses hashes that are spent from  - but then that gives us an oxymoron, as all addresses with sending transactions have valid points. So it's not very useful. As @gmaxwell wrote, probabilistic pruning based on alphanumerical patterns in the pubkey or address could lead to a frozen funds situation for coincidental legit addresses. However, an even lighter version of the Bong Daplus software can cache all UTXOs, exclude such addresses from the cache, and then query a full node when a cache-miss occurs. Because it is backed by full nodes, a fork will never occur.