[clang] [flang] [flang] Add -fdisable-real-16 option (PR #221904)

[Shunsuke Watanabe via cfe-commits]

s-watanabe314 wrote:

Related to #221907.
#221907 makes `REAL(16)` available whenever the backend supports fp128, independently of the availability of quadruple-precision math intrinsic libraries. In such configurations, programs like

```Fortran
integer, parameter :: k = merge(16, 4, selected_real_kind(p=33) .eq. 16)
real(kind=k) :: x
test = sin(x)
```

may select `REAL(16)` and later fail to link if the required quadruple-precision math library support is unavailable. This patch introduces `-fdisable-real-16` as a way to disable `REAL(16)` support and avoid such link-time dependencies.

https://github.com/llvm/llvm-project/pull/221904