smax_from_valkey: --merge option, error handling, scripts README

[mpound]

Summary

• smax_from_valkey.py error handling: Variables whose smax_pull query fails are now omitted from output (previously written as {"size": "?", "smax_type": "?", "_error": "..."} which would pollute smax.json). Failure count reported to stderr; --verbose lists the names.
• smax_from_valkey.py --merge: New option deep-merges generated entries into an existing smax.json. smax.json values take precedence on conflicts, preserving validated descriptions, units, and thresholds while adding new keys. Output goes to --output.
• scripts/README.md: New reference doc covering all scripts — dump_redis, compare_monitor_points, smax_from_valkey (including merge workflow), fakeobs, simple_smax_check, and test files.
• recursewithtreelib.py: Fix broken import paths (Marc).
• 9 new TestDeepMerge tests and 5 new TestBuildSubtreeErrors tests. Total: 223 tests passing.

Typical workflow enabled by --merge

# 1. Dump all Valkey leaves
uv run scripts/dump_redis.py --flat --strip > docs/all_monitor_points.txt

# 2. Find what is missing from smax.json
uv run scripts/compare_monitor_points.py --missing > valkeynotsmax.txt

# 3. Generate entries and merge directly into smax.json
uv run scripts/smax_from_valkey.py valkeynotsmax.txt \
    --merge --output smax_updated.json --verbose

# 4. Review, add descriptions/units, replace conf/smax.json

Test plan

• uv run pytest — 223 tests pass
• uv run scripts/smax_from_valkey.py valkeynotsmax.txt --merge --output /tmp/test.json --verbose with a live Valkey server

🤖 Generated with Claude Code