feat(regression): add E2E canary regression suite (tests/regression/)

Seven canaries prove both halves of the server's job:
- GREEN: good apps are reported healthy (good-simple + good-significant)
- RED: broken apps are caught at intended tier (false-green guard + 4 per-tier)

Fixtures: custom-html-bkp-bad (backup tier RED) + custom-html-rst-bad (restore tier RED).
All 7 canaries verified on live server (see STATUS-regression.md for artifacts).

Not wired to per-commit CI — run on-demand: pytest -m canary tests/regression/

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

tests/custom-html-bkp-bad/ops.py

@@ -0,0 +1,19 @@
+"""custom-html-bkp-bad — lifecycle ops for bad-backup/bad-restore RED canaries.
+
+Intentionally has NO pre_backup hook: the marker is never seeded before backup,
+so the backup snapshot has no ci-marker.txt. pre_restore writes "mutated" so that if
+restore DOES bring back the snapshot, the marker is gone/still-mutated → test fails.
+"""
+
+from __future__ import annotations
+
+from harness import lifecycle
+
+MARKER_PATH = "/usr/share/nginx/html/ci-marker.txt"
+
+
+def pre_restore(domain: str, meta: dict) -> None:
+    """Write 'mutated' to the marker before restore runs. If restore brings back the
+    snapshot (which has no marker — never seeded by pre_backup), the marker ends up
+    MISSING or 'mutated' after restore → test_restore_returns_state FAILS → restore=RED."""
+    lifecycle.exec_in_app(domain, ["sh", "-c", f"echo mutated > {MARKER_PATH}"])