Fix a dead store in setSPairGroupSize, and delete the unbuilt MESClassicGBAlg

[d-torrance]

Two independent cleanups in ClassicGBAlg's neighbourhood, neither changing
behaviour.

1. setSPairGroupSize assigned the default to its own parameter

void ClassicGBAlg::setSPairGroupSize(unsigned int groupSize) {
  if (groupSize == 0)
    groupSize = mReducer.preferredSetSize();   // goes nowhere
  else
    mSPairGroupSize = groupSize;
}

groupSize is the by-value parameter; mSPairGroupSize is the member the
algorithm reads. So passing 0 — which the CLI documents as "use an appropriate
default" — did nothing at all.

The output was right anyway, because the constructor already initialises
mSPairGroupSize to reducer.preferredSetSize(), the same value the discarded
line computes. The two cannot disagree: preferredSetSize is a constant per
reducer (1 for TypicalReducer, 100000 for F4Reducer) and mReducer is
bound to the object the constructor asked. So the branch that appears to
establish the default was the branch that did nothing, and the default was
correct by coincidence.

The likely origin is a mis-edit of

if (groupSize == 0)
  groupSize = mReducer.preferredSetSize();
mSPairGroupSize = groupSize;

where the trailing assignment became an else, which leaves the nonzero path
working and guts the zero path.

No behaviour change, and none is observable — ClassicGBAlg is declared inside
its own .cpp and has one caller, so there is no seam for a test. Checked by
hand through the value mgb prints as S-pair group size:

	before	after
-sPairGroupSize 0, default reducer	1	1
-sPairGroupSize 0 -reducer 26	100000	100000
-sPairGroupSize 5	5	5

It matters to whoever next changes how that default is chosen, since their edit
to that line would silently have no effect.

2. MESClassicGBAlg is deleted

MESClassicGBAlg.cpp and .hpp arrived in e6996a3, a bulk backport of 14
files from Macaulay2/M2#2160 "Refactor mathic", and have had no substantive
change since.

Never built. Absent from Makefile.am and src/CMakeLists.txt including
EXTRA_DIST, so no release tarball has ever contained them. build/vs12's
project file omitted them too, until PR #78 deleted it. Nothing in the tree
refers to them.

Cannot be built. The .cpp includes ClassicGBAlg.hpp rather than its own
header and defines mgb::ClassicGBAlg — the class ClassicGBAlg.cpp already
defines. Adding it to a build is a duplicate symbol. Its own header declares
something else entirely, mgbF4, which the .cpp never includes.

Nothing in it to keep. The algorithm is identical — computeGrobnerBasis
matches outright, step and insertReducedPoly differ only in spelling
MATHICGB_ASSERT as plain assert. Every other difference is a removal: the
statistics block built from 51 ColumnPrinter calls is gone, printMemoryUse
is an empty body reading // TODO: bring over from mathicgb or rewrite, and
LogDomain is dropped. 226 fewer lines, all subtractions.

It was a port in progress — stripping this project's macros so the file would
compile inside Macaulay2's engine, which its own compile-command names as
$M2BUILDDIR/Macaulay2/e. Macaulay2 does not have it today either; the name
appears nowhere in that repository.

Deleting it also removes a second, untouched copy of the mislabelled timing
line PR #87 fixed, and a third site of an ignored queueType parameter.
Mostly it stops every grep for a bug in ClassicGBAlg.cpp returning two
answers, one of which compiles nowhere. Recoverable from history.

Verification

• 253/253 in cmake Release and Debug, no build warnings from GCC 13.3, at each
  commit.
• The S-pair group size table above, measured before and after.
• No reference to MESClassic, to mgbF4, or to its include guard survives in
  any tracked file.

No autotools distcheck run locally; neither commit touches a build system and
CI covers it.

🤖 Generated with Claude Code