[AArch64][SVE] can fold svmul by -1 to svneg

[k-arrows]

Test case:

#include <arm_sve.h>

svuint64_t foo (svuint64_t x, svbool_t pg)
{
  return svmul_n_u64_x (pg, x, -1);
}

https://godbolt.org/z/GoEnMcc9s

I referred to the following GCC patch.
patch

[llvmbot]

@llvm/issue-subscribers-backend-aarch64

Author: None (k-arrows)

Test case: ```cpp #include <arm_sve.h>

svuint64_t foo (svuint64_t x, svbool_t pg)
{
return svmul_n_u64_x (pg, x, -1);
}

https://godbolt.org/z/GoEnMcc9s

I referred to the following GCC patch.
[patch](https://github.com/gcc-mirror/gcc/commit/f9c99d403c9a0948936e3120ad97b4f10998351f)
</details>

[MartinWehking]

Hi @k-arrows,

Thanks for reporting this. We are looking into it.

[MartinWehking]

Hi @k-arrows ,

We have put up the following PR to address the issue:
#156906

[k-arrows]

Fixed by #160828. I will close this issue.