Implement rule priority ordering and conflict resolution

Implement getPrioritizedMatch(List candidates) as a pure synchronous method — no async, no repository calls. Use Dart's List.sort() with a custom Comparator that applies the three-tier ordering: (1) descending by rule.priority, (2) descending by computeSpecificityScore(rule), (3) ascending by rule.insertionOrder. Extract computeSpecificityScore(ScenarioRule) as a private static method for independent testability. Do not add the specificity score to the ScenarioRule data model — compute it on the fly to keep the model a pure data class.

Update the evaluate() method from task-005 to collect all matching rules into a List candidates, then call getPrioritizedMatch(candidates) instead of returning the first match found. Store candidates in EvaluationResult.candidateRules before applying priority. If ScenarioRule does not yet have a priority field, add it now with a default of 0 and document the convention that higher values mean higher priority.

Write unit tests using flutter_test. Tests must cover: (1) single rule returns immediately, (2) two rules with different explicit priorities, (3) two rules with equal priority but different specificity scores, (4) two rules with equal priority and equal specificity but different insertion orders, (5) all three tiebreakers needed in sequence, (6) empty list returns null, (7) determinism test — shuffle input and assert same winner across 10 random orderings, (8) specificity score calculation for all 16 permutations of the 4 binary condition dimensions. Verify that evaluate() in task-005 correctly pipes all matching rules through getPrioritizedMatch and that EvaluationResult.candidateRules is populated correctly.