Refine active-rule deferral path matching; expand interpreter docs with concrete examples

[Copilot]

This updates the data lookup deferral path in the interpreter to be clearer and more efficient. It adds thorough inline documentation and examples while tightening path matching to avoid unnecessary allocation in hot evaluation flow.

• Deferral matching logic

  • Replaced joined-string matching (module_path + "." + rule_path) with segment-aware prefix checks.
  • Added request_matches_active_rule_path(requested_path, module_path, rule_path) to express matching semantics directly.

• Performance/idiomatic Rust

  • Removed per-iteration format! allocation inside active-rule evaluation checks.
  • Uses strip_prefix + boundary-aware helper logic for allocation-light path comparisons.

• Comments and examples

  • Expanded helper docs with explicit true/false examples for segment-boundary prefix behavior.
  • Clarified dynamic lookup semantics (data.a.b and data.a.b. descendants) and recursion wording for cross-package/sibling re-entry.

// Before (allocates each iteration)
let full_rule_path = format!("{}.{}", module_path, rule_path);
if Self::path_is_prefix(requested_path, &full_rule_path) { ... }

// After (allocation-free matching)
if Self::request_matches_active_rule_path(requested_path, &module_path, &rule_path) { ... }

[anakrish]

Document this nicely. Not clear what is going on.

[anakrish]

Is this called in the hot path?