fix(difftest): fix alignment for C++ type declarations

[NewPaulWalker]

When generating arrayType, not all 8-bit-aligned types are valid in C++.

For example, if someone defines a 41-bit-wide signal, it may result in generating an unsupported C++ type like uint48_t instead of a valid uint64_t.

This change ensures that for any given bit width, a C++-supported type is generated.

[klin02]

We have used the same alignment in many other places. If such a situation (like a 41-bit-wide signal) does exist, we will make the changes together in this PR.

[NewPaulWalker]

We have used the same alignment in many other places. If such a situation (like a 41-bit-wide signal) does exist, we will make the changes together in this PR.

It's OK! Should I close this PR directly now, or temporarily convert it to a draft?

[poemonsense]

We have used the same alignment in many other places. If such a situation (like a 41-bit-wide signal) does exist, we will make the changes together in this PR.

How is this PR going?

[klin02]

How is this PR going?

I think if necessary, we can replace all (size + 7) / 8 * 8 with a common method, or simply leave it as it is.

[poemonsense]

How is this PR going?

I think if necessary, we can replace all (size + 7) / 8 * 8 with a common method, or simply leave it as it is.

We need a conclusion for each PR whose author has done his/her work.

Either we, as maintainers, will accept it in the future or now (then do it ASAP) or not accept it (and close it)

[klin02]

How is this PR going?

I think if necessary, we can replace all (size + 7) / 8 * 8 with a common method, or simply leave it as it is.

We need a conclusion for each PR whose author has done his/her work.

Either we, as maintainers, will accept it in the future or now (then do it ASAP) or not accept it (and close it)

OK. I will try to find other similar alignment issue, and fix them ASAP.

[klin02]

This update refines the alignment for difftest elems. Previously, element widths
were simply rounded up to the nearest multiple of 8 bits, which could produce
unsupported C++ types such as uint48_t (e.g., for a 41-bit signal). We now align
each element to the nearest supported C++ type (uint8_t, uint16_t, uint32_t,
uint64_t), ensuring correct code generation and type safety in the C++ logic.