Remove the BlueGene compilers from the specification

[bbiiggppiigg]

Title: Remove the BlueGene compilers from the specification

Split out of #276 per review — the spec fix that PR needed to regenerate the
platform directories, extended to remove the BlueGene/Q compilers as
requested.

Commit 1 — spec: drop optimization translations for removed bluegene-P compilers

The bg_gcc, bg_g++, and bg_gfortran compilers no longer have any
platform (the bluegenep platforms were removed from the specification some
time ago), but they still had compiler_opt_trans facts. That trips the
"not defined as a compiler, but has optimization translation defined"
sanity check in write_tuples, so gprolog halts before writing any tuples —
making it impossible to regenerate the platform directories from the current
specification.

Commit 2 — spec: remove the BlueGene compilers

Removes the eight BlueGene compiler definitions and every fact referencing
them:

• BlueGene/P: bg_gcc, bg_g++, bg_gfortran (already unreachable, see
  above)
• BlueGene/Q: bgq_gcc, bgq_g++, bgq_gfortran, bgxlc, bgxlc++

Covers compiler_format, compiler_platform, mutatee_comp,
mutator_comp, compiler_define_string, compiler_s, compiler_pic,
compiler_static_link/compiler_dynamic_link, comp_std_flags_str,
comp_mutatee_flags_str, mutatee_link_options, the comp_lang /
compiler_opt_trans / compiler_pic_trans / compiler_parm_trans /
compiler_for_mutatee membership lists, and the bgxlc-only -qpic
translation clause.

Commit 3 — spec: remove the BlueGene/Q platform and the bluegene OS guards

With the compilers gone, ppc64_bgq_ion is fully inert, so this removes the
platform fact, platform_abi, the platform-list entry, and every rule keyed
on the 'bluegene' OS (membership guards, \+ platform exclusions, the
bluegene clauses of platform_module/pcPlatforms/runmode_platform, the
gcc/g++ OS \= 'bluegene' conditions, and the BG/P platform_mode
VN/DUAL/SMP machinery). remote_platform/1 and
mutatee_compiler_platform_exclude/2 had only bluegene clauses but are still
consulted during generation and are declared discontiguous (not dynamic), so
each keeps a single explicit always-fail clause. After this commit the
specification is entirely BlueGene-free.

Regeneration: platform directories regenerated with gprolog 1.5.0 +
python2 (Makefile.genspec gen-clean gen-all). The only generated-file
change is the bgq compiler metadata disappearing from each platform's
tuples.gz; no mutatee/mutator build lists change on any checked-in
platform. (Note: the checked-in tuples.gz were stale relative to the
current spec — they still carried the removed
ppc32_linux/ppc32_bgp*/ppc64_bgq* platform metadata — so regenerating
rewrites them; all harness-driving generated files reproduce byte-for-byte.)

#276 will be restacked on this branch and shrinks to just the ported-test
removal.

[bbiiggppiigg]

@hainest do you want me to remove all reference to these compilers transitively?

[hainest]

@hainest do you want me to remove all reference to these compilers transitively?

Yeah. They are long dead, so we can completely scrub them.

[bbiiggppiigg]

@hainest do you want me to remove all reference to these compilers transitively?

Yeah. They are long dead, so we can completely scrub them.

@hainest done.

[hainest]

LGTM.