Bsb/py catch up

[bsbodden]

Catch-up with recent redis-vl-python changes to maintain feature parity.

• Configurable max_distance in MultiVectorQuery — Replace hardcoded DISTANCE_THRESHOLD=2.0 with per-Vector maxDistance field (validated 0.0–2.0). Changes range query join from | to AND to match
  upstream. (Python commit 61f0f41)
• ChatRole enum and count() for message history — Add ChatRole enum (USER, ASSISTANT, SYSTEM, TOOL) as single source of truth for role validation, with backward-compatible acceptance of deprecated
  "llm" role. Add count()/count(sessionTag) to MessageHistory and SemanticMessageHistory using CountQuery. (Python commits 23ecc77, e7301a2)
• toString() on core public classes — Add human-readable toString() to SearchIndex, SemanticCache, LangCacheSemanticCache, SemanticRouter, MessageHistory, and SemanticMessageHistory. (Python commit
  e285fed)

---

Note

Medium Risk
Changes multi-vector search query construction (per-vector maxDistance and AND-joined range clauses), which can alter result sets and performance. Also adjusts message-history role validation and adds new counting APIs, which may affect callers relying on previous role strings or behavior.

Overview
Catches up to recent redis-vl-python parity by changing multi-vector aggregation queries to use per-vector maxDistance (validated 0.0–2.0) instead of a hardcoded threshold, and by joining VECTOR_RANGE clauses with AND rather than |.

Adds a ChatRole enum for message-history role validation (accepting deprecated llm with a warning) and introduces count()/count(sessionTag) on message histories via CountQuery.

Adds human-readable toString() implementations for several public classes (SearchIndex, SemanticCache, LangCacheSemanticCache, SemanticRouter, MessageHistory, SemanticMessageHistory) with new unit/integration coverage for toString, role validation, message counts, and max-distance behavior.

^{Written by Cursor Bugbot for commit b983a9c. This will update automatically on new commits. Configure here.}

[jit-ci]

🛡️ Jit Security Scan Results

✅ No security findings were detected in this PR

---

^{Security scan by Jit}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, have a team admin enable autofix in the Cursor dashboard.}

[cursor]

Scientific notation in query string for small maxDistance

Medium Severity

Using %s to format the double maxDistance into the Redis query string calls Double.toString(), which produces scientific notation (e.g., "5.0E-4") for values below 0.001. Since the validation allows any value in [0.0, 2.0], a user setting maxDistance(0.0005) would generate a VECTOR_RANGE 5.0E-4 query fragment that Redis likely cannot parse, causing a query failure. The previous code used %.1f which always produced plain decimal output. A format like BigDecimal.valueOf(...).toPlainString() would preserve precision without risking scientific notation.