Followup to d72826a5df. Jens reports
in the bug that he sees a crash w/ GCC 15 in my_convert too. It's very
likely the same problem: 15 at least already had some peeling for alignment
changes and we know the code in MariaDB is broken here.
Don't bother conditionalising it on GCC as Clang accepts -fno-tree-vectorize
too and it could show up there too of course as it's UB.
And revbump here as well given it clearly affects real users, not just
those w/ experimental (16) GCC.
Closes: https://bugs.gentoo.org/959423
Signed-off-by: Sam James <sam@gentoo.org>
8196f16d2f