[cc] Document [u]intmax_t

[lenary]

These have some effect on the ABI of code that use them, and need to be kept stable long-term, so we should document the sizes they are.

[kito-cheng]

I am little hesitate to document that down in the ABI spec, do you mind give few more background about why we need to document that? and the follow up question is: should we also document all other [u]int*_t from the C language?

[lenary]

Here is more background on how this leaks into the ABI:

• https://thephd.dev/intmax_t-hell-c++-c
• https://www.open-std.org/jtc1/sc22/wg14/www/docs/n2425.pdf

I have actually opened riscv-non-isa/riscv-c-api-doc#129 to not document the other int_fastN_t and int_leastN_t in the ABI as they currently don't always match between toolchains, and I don't think we want to make them ABI stable. Sorry for not cross-linking to that change originally.

[kito-cheng]

We've discussed this topic in last psABI meeting, and we agree defined that in psABI will help RISC-V has more stable ABI even the C spec isn't mandatory it.