[mlir][arith] doc updates for ub semantics, and int representations (#72932)

Following the discussions in this thread,
https://discourse.llvm.org/t/some-question-on-the-semantics-of-the-arith-dialect/74861

,
here are some updates to the documented semantics of Arith.
Added are clarifications on poison behaviour, UBs, overflow semantics,
and the underlying two's complement representation used for integers

Co-authored-by: kuhar <jakubk@openxla.org>
Co-authored-by: math-fehr <mathieu.fehr@gmail.com>