[FXML-4448] Add EmitC lowering for arith.{trunci,extsi,extui}

[cferry-AMD]

This PR introduces EmitC lowering for various integer casts in the Arith dialect.

Per C++ Reference:

• If the destination type is unsigned, the resulting value is the smallest unsigned value equal to the source value modulo 2^n where n is the number of bits used to represent the destination type. That is, depending on whether the destination type is wider or narrower, signed integers are sign-extended or truncated and unsigned integers are zero-extended or truncated respectively.

• If the destination type is signed, the value does not change if the source integer can be represented in the destination type. Otherwise the result is

  • implementation-defined (until C++20)
  • the unique value of the destination type equal to the source value modulo 2^n where n is the number of bits used to represent the destination type(since C++20)
    (note that this is different from signed integer arithmetic overflow, which is undefined).

So, truncation is only guaranteed if unsigned casts are performed, or C++20 is used. This implementation sticks to unsigned where we know it is a truncation, like trunci, and index_cast with greater source width than destination. The only issue is when using index types (that don't have a bitwidth) and C++20 isn't used -- the resulting behavior is then implementation-defined.

[josel-amd]

Small comments here and there

[mgehre-amd]

How about removing the index casts from here and open a separate PR for them once the size_t/ssize_t types are available?

[cferry-AMD]

@mgehre-amd, that's a good way forward.

[mgehre-amd]

Looks good!