fix: defeasible reason codes, _subjectAsObject unary semantics, checker reason whitelist

- LogicalOperators normal mode now emits top-level reason codes
  (never_rule_triggered / requirements_not_met / defeated_by_unless) when a
  defeasible rule resolves to 0, instead of degrading to no_matching_rule.
- RuleEvaluator honors rule._subjectAsObject: unary DSL predicate calls inside
  binary evidence (banned(user) within can_open(user, doc)) check the relation
  on the subject itself, matching the unary fact's self-edge.
- AuthorizationChecker reason whitelist now preserves the defeasible reason
  codes so the checker reports why a defeasible rule denied.

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@arbiter/core",
-  "version": "1.0.1",
+  "version": "1.0.2",
   "description": "Arbiter core engine: graph indices, relation/reachability, authorization rule evaluator, DSL/AST, condensed & sharded snapshots, and evidence fusion.",
   "license": "ISC",
   "author": "",

src/authorization/AuthorizationChecker.js

@@ -507,6 +507,12 @@ export class AuthorizationChecker {
       if (res.reason === 'values_compared_comparison_true') reason = 'values_compared_comparison_true';
       if (res.reason === 'values_compared_comparison_false') reason = 'values_compared_comparison_false';
       if (res.reason === 'values_compared_comparison_insufficient') reason = 'values_compared_comparison_insufficient';
+      // Add defeasible logic reasons (normal mode surfaces these when a
+      // defeasible rule resolves to 0 — never, requirements, or defeaters).
+      if (res.reason === 'defeated_by_unless') reason = 'defeated_by_unless';
+      if (res.reason === 'never_rule_triggered') reason = 'never_rule_triggered';
+      if (res.reason === 'requirements_not_met') reason = 'requirements_not_met';
+      if (res.reason === 'strict_rule_failed') reason = 'strict_rule_failed';
       
        if (resAllowPossibility > maxAllow) {
          maxAllow = resAllowPossibility;

src/authorization/RuleEvaluator.js

@@ -34,7 +34,17 @@ export class RuleEvaluator {
     
     // Convert string keys to numeric IDs if needed
     const numericUserId = typeof userId === 'string' ? this.arbiter.resolveNodeId(userId, options) : userId;
-    const numericObjectId = typeof objectId === 'string' ? this.arbiter.resolveNodeId(objectId, options) : objectId;
+    let numericObjectId = typeof objectId === 'string' ? this.arbiter.resolveNodeId(objectId, options) : objectId;
+    
+    // Unary / subject-scoped rules (e.g. a DSL predicate call `banned(user)`
+    // inside a binary evidence) check the relation on the SUBJECT itself — the
+    // object is the user. The DSL generator marks these with _subjectAsObject;
+    // the evaluator rewrites the object to the subject so (u, banned, u) matches
+    // the unary fact's self-edge instead of (u, banned, object).
+    if (rule._subjectAsObject) {
+      numericObjectId = numericUserId;
+      objectKey = userKey;
+    }
     
     const needsValueContext = valueContext !== null && valueContext !== undefined
       ? true

src/authorization/rules/LogicalOperators.js

@@ -805,6 +805,7 @@ export class LogicalOperators extends BaseRule {
       return {
         possibility: 0,
         reliability,
+        reason: 'never_rule_triggered',
         collectedValues: allCollectedValues,
         meta: {
           ...resultMeta,
@@ -857,10 +858,28 @@ export class LogicalOperators extends BaseRule {
       }
     }
 
+    const finalPossibility = Math.max(0, Math.min(1, possibility));
+
+    // Surface a top-level reason so the AuthorizationChecker reports why a
+    // defeasible rule resolved to 0 instead of a generic 'no_matching_rule'.
+    // Ordering mirrors precedence: requirements beat defeaters; a defeater
+    // reason is only claimed when the base (when/strict) leg was actually
+    // active — otherwise the rule simply did not match.
+    let reason;
+    if (finalPossibility === 0) {
+      const baseWasActive = (defeasibleResult?.possibility || 0) > 0 || (strictResult?.possibility || 0) > 0;
+      if (requiresResult && requiresResult.possibility < 0.5) {
+        reason = 'requirements_not_met';
+      } else if (defeatersResult && defeatersResult.possibility > 0.5 && baseWasActive) {
+        reason = 'defeated_by_unless';
+      }
+    }
+
     return {
-      possibility: Math.max(0, Math.min(1, possibility)),
+      possibility: finalPossibility,
       reliability,
-      validity: buildValidity('product', [], [], 2, Math.max(0, Math.min(1, possibility))),
+      reason,
+      validity: buildValidity('product', [], [], 2, finalPossibility),
       collectedValues: allCollectedValues,
       meta: {
         ...resultMeta,