Add Turin TCB_VERSION support

[deeglaze]

The January 2025 revision of the SEV-SNP specification has more changes than what PR#148 added support for.

The TCB_VERSION structure on EPYC Turin machines is different than Genoa and Milan. We need the abi interpretation of the TCB_VERSION to be product-sensitive.

[deeglaze]

@katexochen Not sure if this is something you're dependent on. I can't prioritize this for a bit since we don't have SNP support deployed on C4D (Turin) machines.

[deeglaze]

Will add a note so I don't forget: the hwID for the KDS query is now also shortened to account for the reduction to 8 octets/bytes. I've asked AMD for clarity about which part of the CHIP_ID field is expected to contain the relevant 8 bytes. I suspect the first 8.

Also there's an apparent typo that the ucodeSPL argument for Turin and beyond should be written "ucodSPL" in the query. Given that I don't have a Turin machine from which to test, I cannot verify if this is indeed the parameter name. This is also part of my query to AMD.