fix(makemaker): write Build proxy when Build.PL is present

[fglock]

Summary

Fix jcpan -t/-i Data::Merger (and any other CPAN dist that ships a Module::Install–style stub Build.PL) failing with:

Alert: no Build file available for 'make ...'
Error: Unable to read file Build
  /jperl Build -- NOT OK

Root cause

Data::Utilities (which contains Data::Merger) ships a Build.PL whose only job is:

# Dear Distribution Packager. ...
require 'Makefile.PL';

CPAN.pm sees Build.PL, sets $self->{modulebuild} = 1, and from then on runs ./Build, ./Build test, ./Build install instead of make .... Our ExtUtils::MakeMaker shim only emitted a Makefile, so ./Build did not exist and the install aborted.

Fix

When WriteMakefile runs and detects a Build.PL in the cwd, also emit a thin ./Build perl script that proxies Module::Build subcommands (build, test, install, clean, realclean, distclean, config) to the matching make targets that _create_install_makefile already generates. Unknown subcommands pass through verbatim.

Touches only src/main/perl/lib/ExtUtils/MakeMaker.pm (plus the auto-injected Configuration.java commit id).

Test plan

• make (full unit suite) passes
• Clean-tree end-to-end run, all phases driven by CPAN.pm:

  rm -rf ~/.cpan/build/Data-Utilities-0.04-*
  ./jcpan -i Data::Merger
  

  • Build.PL -- OK
  • Build -- OK (5 files staged in ./blib)
  • Build test -- OK (Files=7, Tests=55, all pass)
  • Build install -- OK (5 files copied to ~/.perlonjava/lib)
• Runtime smoke test: ./jperl -e 'use Data::Merger qw(merger); ...' merges hashes as expected.

No automated test was added — there are currently no tests in src/test/ for the MakeMaker shim or the jcpan flow to extend. Happy to add one (either a .t that exercises _create_build_script_proxy directly, or a make test-bundled-modules entry that runs jcpan -t Data::Merger) in a follow-up if desired.

Generated with Devin