[Bennyjitsu]

Why

The --author flag on wt merge/wt step commit/wt step squash was added
to forward explicit commit authorship (our own policy: an AI agent must never
commit under the default identity). It turned out to be unnecessary: git
already reads GIT_AUTHOR_NAME/GIT_AUTHOR_EMAIL from the environment on
every code path this flag touched — git commit, git commit-tree (the
--no-ff merge path), and squash's git commit — since nothing in wt
scrubs those vars from spawned children.

Verified live before removing: exported the env vars, ran wt step commit,
wt step squash, and wt merge --no-ff against a real build with the flag,
and confirmed each resulting commit's author was set correctly with the
committer left ambient — the same split the flag's own docs promised.

The flag also had a real bug: split_author_spec didn't match git commit --author's actual semantics (rejected pattern-lookup and Name <> forms
git accepts), and the same --author value was silently accepted on the
squash/commit path but rejected on the --no-ff commit-tree path — worth
fixing, but not worth fixing in code we don't need.

What changed

• Removed the --author CLI flag from merge, step commit, step squash.
• Removed split_author_spec and run_command_with_env (only consumer of
  each was this flag).
• test_step_squash_failed_commit_restores_original_history loses its
  invalid---author failure injector (that path no longer exists) — a real
  .git/hooks/pre-commit exiting non-zero replaces it, firing at the same
  point (after the soft reset, since the squash commit doesn't pass
  --no-verify) and proving the same rollback behavior.
• Removed the four --author-specific tests.
• Regenerated docs, skill reference pages, and the Unix-only help snapshots
  (help.rs is #![cfg(not(windows))], regenerated via WSL2).

Full pre-merge gate green: cargo run -- hook pre-merge --yes — 4,934 tests
passed, clippy clean, doc-sync clean.

🤖 Generated with Claude Code