fix(cli): detect-changes/visualize/wiki/watch no longer create empty graph.db

[arimu1]

Summary

Follow-up to #777 / #782: status already resolved the DB path read-only when the graph was missing, but detect-changes, visualize, wiki, and watch still called get_db_path() (create) + GraphStore() and could materialize an empty schema graph.db. Later incremental updates then only re-parsed changed files, leaving the graph incomplete.

Changes

• Treat those commands like status: resolve DB path with read_only=True (or explicit --data-dir without registry/mkdir side effects).
• Exit with No graph found … Run code-review-graph build first. when the DB is missing (same message as status / dead-code / forget).
• Keep the default-path legacy .code-review-graph.db migration for these commands.
• visualize / wiki still write exports next to an explicit --data-dir when a graph is present, without forcing a registry write when the graph is absent.
• Document recovery for already-poisoned graphs: full code-review-graph build (see docs/TROUBLESHOOTING.md). Auto-heal heuristics for incomplete non-empty graphs left out of scope (issue gap 2).

Test plan

• uv run pytest tests/test_cli_reconciliation.py tests/test_cli.py::test_visualize_json_uses_local_export tests/test_cli.py::TestWatchInteraction -k "detect_changes or missing_graph or read_only" -q
• Parametrized tests: missing graph for status/detect-changes/visualize/wiki/watch does not create data dir or empty graph.db
• Explicit --data-dir for visualize/wiki/watch/status is read-only when missing
• Existing status / visualize / watch regressions still green
• ruff check on touched Python files

Fixes #803

[github-actions]

code-review-graph review

Overall risk: 0.35 (LOW) — 4 changed function(s)/class(es), 0 affected flow(s), 1 test gap(s)

Risk-scored changes

Risk	Level	Symbol	Location	Tested
0.35	low	code_review_graph/cli.py::main	code_review_graph/cli.py:603	no
0.30	low	tests/test_cli.py::test_visualize_json_uses_local_export	tests/test_cli.py:105	(test)
0.30	low	tests/test_cli_reconciliation.py::test_read_only_commands_missing_graph_do_not_create_empty_db	tests/test_cli_reconciliation.py:157	(test)
0.30	low	tests/test_cli_reconciliation.py::test_read_only_commands_data_dir_option_is_read_only	tests/test_cli_reconciliation.py:196	(test)

Test gaps

• code_review_graph/cli.py::main (code_review_graph/cli.py:603)

Token savings: this graph-backed report used ~33,819 fewer tokens (~97%) than reading every changed file in full (estimated, chars/4 approximation).

---

Powered by code-review-graph — local-first analysis; no code leaves the CI runner.