Doesn't that require both coins using the same elliptic curve?
1. A lot of altcoins just copy-pasted secp256k1, so it is not a big deal.
2. Read about DLEQ proofs. It is possible to prove, that the same private key was used on two completely different curves, and then execute a contract in that way.
We could have that decentralized DNS with OP_CSFS, but I cannot think of any other ways.
There are a lot of other ways. The simplest example, which is used even today, is related to vanity addresses: you have a regular Bitcoin address, but you can mine N characters in the name, and have some unique identifier. Even better: imagine what would happen, if you would mine some Silent Payment address, to avoid address reuse.
And that way is not only limited to Bitcoin: people also mine *.onion domains, in exactly the same way.
Do you understand what the OP is talking about?
As far as I know, a "basic generic Elliptic Curve Point Contract functionality" is what you will have, if you introduce OP_CHECKSIGFROMSTACK. But yes, you can have "<pubkey> <pubkey> OP_PUBKEY_ADD" or "<number> <pubkey> OP_PUBKEY_MUL" instead (or even "<numberAdd> <numberMul> <pubkey> OP_PUBKEY_MUL_WITH_ADD"). But in general, a signature is "multiplication and addition". Which means, that if you can sign any given message, then you unlock "mul and add" functionality, just by packing those two 256-bit numbers as a single signature.