feat(8.0): brain.fillSubtypes migration helper + pre-RC1 gap closure

- brain.fillSubtypes(rules): idempotent subtype back-fill for pre-8.0 data.
  One rule per NounType/VerbType (literal default or per-entry function);
  fills only entries still missing a subtype through the public update()/
  updateRelation() paths; returns { scanned, filled, skipped, errors, byType }.
  Full unit suite in tests/unit/brainy/fill-subtypes.test.ts.
- Fix getNouns/getVerbs pagination hasMore (peek one past the window) —
  was permanently false, silently truncating every multi-page walk.
- find({ near }) without near.id now throws a teaching error instead of an
  opaque storage sharding failure; CLI --threshold without --near applies a
  plain score floor.
- CLI init/close audit: every one-shot command init()s, close()s, and exits
  explicitly; delete the unmaintained interactive REPL; replace the cloud-era
  storage subcommands with status/batch-delete; new types/validate commands.
- requireSubtype JSDoc now documents the 8.0 default-on contract; audit()
  recommendation points at fillSubtypes.
- Docs: data-storage-architecture rewritten to the real 8.0 on-disk layout;
  README storage section reflects filesystem+memory and snapshots; eli5 and
  SEMANTIC_VFS /as-of/ semantics corrected; internal tracker IDs and
  .strategy references scrubbed from published files.

src/storage/baseStorage.ts

@@ -1528,10 +1528,16 @@ export abstract class BaseStorage extends BaseStorageAdapter {
 
     const { limit, offset = 0, filter } = options
     const collectedNouns: HNSWNounWithMetadata[] = []
+    // Collect ONE item past the requested window so `hasMore` is decidable:
+    // stopping exactly at offset+limit cannot distinguish "page full, nothing
+    // after it" from "page full, more behind it" — which made hasMore
+    // permanently false and silently truncated every multi-page walk
+    // (audit/migrateField/fillSubtypes, graph verb-id recovery, count rebuilds).
     const targetCount = offset + limit
+    const peekCount = targetCount + 1
 
     // Iterate by shards (0x00-0xFF) instead of types
-    for (let shard = 0; shard < 256 && collectedNouns.length < targetCount; shard++) {
+    for (let shard = 0; shard < 256 && collectedNouns.length < peekCount; shard++) {
       const shardHex = shard.toString(16).padStart(2, '0')
       const shardDir = `entities/nouns/${shardHex}`
 
@@ -1539,7 +1545,7 @@ export abstract class BaseStorage extends BaseStorageAdapter {
         const nounFiles = await this.listCanonicalObjects(shardDir)
 
         for (const nounPath of nounFiles) {
-          if (collectedNouns.length >= targetCount) break
+          if (collectedNouns.length >= peekCount) break
           if (!nounPath.includes('/vectors.json')) continue
 
           try {
@@ -1597,7 +1603,8 @@ export abstract class BaseStorage extends BaseStorageAdapter {
       }
     }
 
-    // Apply pagination
+    // Apply pagination. The peeked extra item (if any) is dropped by the slice;
+    // its existence is exactly what makes hasMore true.
     const paginatedNouns = collectedNouns.slice(offset, offset + limit)
     const hasMore = collectedNouns.length > targetCount
 
@@ -1647,7 +1654,11 @@ export abstract class BaseStorage extends BaseStorageAdapter {
 
     const { limit, offset = 0, filter } = options  // cursor intentionally not extracted (not yet implemented)
     const collectedVerbs: HNSWVerbWithMetadata[] = []
-    const targetCount = offset + limit  // Early termination target
+    // Same peek-one-past-the-window strategy as getNounsWithPagination — see
+    // the comment there. Without the extra item, hasMore is undecidable and
+    // was permanently false (silent truncation of every multi-page walk).
+    const targetCount = offset + limit  // Requested window end
+    const peekCount = targetCount + 1   // Early termination target (window + 1 peek)
 
     // Prepare filter sets for efficient lookup
     const filterVerbTypes = filter?.verbType
@@ -1668,7 +1679,7 @@ export abstract class BaseStorage extends BaseStorageAdapter {
       : null
 
     // Iterate by shards (0x00-0xFF) instead of types - single pass!
-    for (let shard = 0; shard < 256 && collectedVerbs.length < targetCount; shard++) {
+    for (let shard = 0; shard < 256 && collectedVerbs.length < peekCount; shard++) {
       const shardHex = shard.toString(16).padStart(2, '0')
       const shardDir = `entities/verbs/${shardHex}`
 
@@ -1676,7 +1687,7 @@ export abstract class BaseStorage extends BaseStorageAdapter {
         const verbFiles = await this.listCanonicalObjects(shardDir)
 
         for (const verbPath of verbFiles) {
-          if (collectedVerbs.length >= targetCount) break
+          if (collectedVerbs.length >= peekCount) break
           if (!verbPath.includes('/vectors.json')) continue
 
           try {
@@ -1737,9 +1748,11 @@ export abstract class BaseStorage extends BaseStorageAdapter {
       }
     }
 
-    // Apply pagination (Efficient slicing after early termination)
+    // Apply pagination. The peeked extra item (if any) is dropped by the slice;
+    // its existence is exactly what makes hasMore true. `>` (not `>=`) keeps the
+    // exact-boundary case (total == offset+limit) from looping forever.
     const paginatedVerbs = collectedVerbs.slice(offset, offset + limit)
-    const hasMore = collectedVerbs.length > targetCount  // Fixed >= to > (was causing infinite loop)
+    const hasMore = collectedVerbs.length > targetCount
 
     return {
       items: paginatedVerbs,
@@ -3235,8 +3248,8 @@ export abstract class BaseStorage extends BaseStorageAdapter {
    * Before this override the Uint32Array counters were only persisted
    * on a heuristic schedule inside `saveNoun_internal` (first-of-type or
    * every-100th), which left readers seeing stale counts after a clean
-   * writer flush — the same failure mode that BR-FIND-WHERE-ZERO surfaced
-   * via `brain.stats()`.
+   * writer flush — the same silent-stale failure mode that once produced
+   * zero counts from `brain.stats()`.
    */
   public async flushCounts(): Promise<void> {
     await super.flushCounts()