Followups to int128_struct arithmetic

[sipa]

This is a follow-up to #1000:

• Add randomized unit tests for int128 logic.
• Add CI for the _(u)mulh code path (on non-ARM64 MSVC).
• Add heuristic logic to enable int128_struct based arithmetic on 64-bit MSVC, or systems with pointers wider than 32 bits.
• Fix signed overflow in ARM64 MSVC code.

[roconnor-blockstream]

why not ((int128_t)hi << 64) + lo?

[sipa]

((int128_t)hi << 64) + lo is UB if hi is negative (left-shift of negative value is undefined).

((uint128_t)hi << 64) + lo would work, but involves sign-extending hi, and then throwing those extension bits away.

((uint128_t)(uint64_t)hi << 64) + lo is equivalent, but avoids the sign-extension, which I thought would be more obviously correct.

[roconnor-blockstream]

Somehow I missed the fact the left-shift of negative values is UB. In retrospect, that makes a bit of sense if you cannot count on two's complement, even if the formal definition of multiplying by 2^n still makes sense for negative numbers.

[real-or-random]

utACK 3afce0a

[roconnor-blockstream]

utACK 99bd335

[real-or-random]

utACK 99bd335

[jonasnick]

utACK 99bd335

EDIT: Maybe #1158 (comment) should be merged first?

[real-or-random]

ACK 99bd335 tested this also on MSVC locally with the override, including all the benchmark binaries

EDIT: Maybe #1158 (comment) should be merged first?

I think we should be greedy and merge it now that it has enough ACKs. The other one will be rather easy to rebase.