fix: silence the GCC doc build (no texi2pod.pl in this tree)

[mgrossmann]

Problem

Every clean make / make compiler printed six of these pairs:

perl ../../cc370/gcc/../contrib/texi2pod.pl .../doc/invoke.texi > gcc.pod
Can't open perl script ".../contrib/texi2pod.pl": No such file or directory
make[1]: [gcc.pod] Error 2 (ignored)
pod2man: unable to format gcc.pod
make[1]: [doc/gcc.1] Error 1 (ignored)

4077723 ("strip non-build cruft from the GCC tree") removed contrib/, and with it contrib/texi2pod.pl — but not the target that consumes it. configure finds Pod::Man >= 1.10, sets GENERATED_MANPAGES=generated-manpages, and doc: then walks doc/{gcc,cpp,gcov}.1 + doc/{gfdl,gpl,fsf-funding}.7, failing on each and leaving a 10-byte timestamp stub in build/gcc/doc/. The errors are --prefixed, so the build succeeded — it was pure noise.

Fix

Hardcode both doc knobs empty in cc370/gcc/Makefile.in instead of restoring the script:

• GENERATED_MANPAGES — nothing installs those six pages. install-man ships only $(MAN1), generated from this repo's own man/*.pod, and there is no gcc/cpp/gcov binary to document. Restoring texi2pod.pl would add six texi→pod→man conversions per clean build for files nobody uses.
• BUILD_INFO — empty today only because the dev host has no makeinfo. On a host with texinfo installed, configure sets it to info and pushes GCC 3.4.6 texinfo through a modern makeinfo for info files we equally do not ship. Hardcoding makes the build host-independent rather than accidentally quiet.

doc: is now just gccbug.

Verification

Fresh out-of-tree build (configure + all-gcc) on a clean tree:

• no texi2pod / pod2man lines in the log, no Error … (ignored)
• build/gcc/doc/ stays empty — no timestamp stubs
• cc1 + xgcc build; -dumpmachine → cc370; -O1 -S on a trivial TU emits the expected COPY PDPTOP / GCCMVS!! prologue
• cc370/tests/run.sh → ALL CC370 TESTS PASSED

Unaffected and still present: the configure-time warning *** Makeinfo is missing or too old. *** Info documentation will not be built. — that comes from configure's own check, not from the doc rules.