The `tcache_fiber_migration` reproducer only exercises the bug when the
allocator is statically linked and inlined under whole-program LTO. None of the
existing lanes build that way, so add a dedicated Linux lane: clang ThinLTO,
`--with-jemalloc-prefix=je_`, and llvm-ar/nm/ranlib (to archive the LTO bitcode)
+ `-fuse-ld=lld`, then `make check`.
Authored in scripts/gen_gh_actions.py; .github/workflows/linux-ci.yml
regenerated.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>