strchr, PR 2767, c_std vs c_shadow, and branch failure

[Benjamin Kosnik]

> Basically, it's PR 2767.  Linking anything against the shared lib gives
> "undefined reference to `std::strchr(char const*, int)'".

? my mkcheck results don't jibe with this

> Switching to c_shadow as noted in the PR trail led to more problems (see
> the "abort() in c_shadow?" email), which finally proved too confusing
> for me (no declarations of size_t visible to anything, anywhere, since
> gcc/include/stddef.h gets confused apparently).  The move to c_shadow is
> still the right way to go, but I need advice before I go further.

c_shadow is broken, bit rotted a bit