[Patch, libgfortran] Fix PR23154 Currupt string and segfault informattedread.

[FX Coudert]

> I suspect the pyCmd and tclCmd failures are due to fortran and C i/o from
> a pipe not being synchronized (despite setting the GFORTRAN_UNBUFFERED_ALL
> environmental). We may not be able to do much about that in the near term.

On the contrary, I am very interested in those. Is the problem with C 
output (that is, if you add a flush() after each C output) or with Fortran?

FX