fix(rag): export MetadataScan types + align Pinecone config contract

IVectorStore.ts was carrying local changes (MetadataScanOptions,
MetadataScanResult, scanByMetadata(), numeric+string $gt/$gte/$lt/$lte
comparison operators) behind a git skip-worktree flag, so the exports
never reached origin. PineconeVectorStore.ts imports MetadataScanOptions
+ MetadataScanResult from '../IVectorStore.js' (line 27-28); CI tsc
failed with TS2305 "has no exported member" on both since the committed
IVectorStore.ts didn't export them.

VectorStoreConfiguration.ts carried a parallel local change that
replaced the hand-defined PineconeVectorStoreConfig (with `environment`
property) with a type-alias to the adapter's real config (with
`indexHost` property). Without this change committed, the Extract<...>
cast at VectorStoreManager.ts:179 fails with TS2345 because the two
PineconeVectorStoreConfig shapes diverge.

Both files were marked skip-worktree, hiding the changes from git push.
This commit clears the skip-worktree state by re-staging + committing
the intended content. CI now has the missing exports and type
alignments.

Changes:
- IVectorStore.ts: add MetadataScanOptions / MetadataScanResult types,
  scanByMetadata() optional method, relax $gt/$gte/$lt/$lte to accept
  string (lexicographic comparison, useful for ISO timestamps).
- VectorStoreConfiguration.ts: re-export PineconeVectorStoreConfig as
  type alias to the adapter's config; switch system default
  defaultProviderId from pinecone_main_prod to qdrant_main_prod +
  update the example data source to match.

Local pnpm run build passes clean after the fix.

Author: jddunn

src/core/config/VectorStoreConfiguration.ts

@@ -14,6 +14,7 @@
 import type { VectorStoreProviderConfig } from '../vector-store/IVectorStore.js';
 import type { StorageAdapter, StorageResolutionOptions } from '@framers/sql-storage-adapter';
 import type { HnswlibVectorStoreConfig } from '../../rag/vector_stores/HnswlibVectorStore.js';
+import type { PineconeVectorStoreConfig as AdapterPineconeVectorStoreConfig } from '../../rag/vector_stores/PineconeVectorStore.js';
 import type { QdrantVectorStoreConfig } from '../../rag/vector_stores/QdrantVectorStore.js';
 import type { Neo4jVectorStoreConfig } from '../../rag/vector_stores/Neo4jVectorStore.js';
 
@@ -103,23 +104,15 @@ export interface LocalFileVectorStoreConfig extends VectorStoreProviderConfig {
 /**
  * Specific configuration for a Pinecone vector store.
  *
- * @interface PineconeVectorStoreConfig
- * @extends VectorStoreProviderConfig
- * @property {'pinecone'} type - Identifier for the Pinecone provider type.
- * @property {string} apiKey - Pinecone API key. Ensure this is securely managed (e.g., via environment variables).
- * @property {string} environment - Pinecone environment (e.g., 'gcp-starter', 'us-east-1-aws').
- * @property {number} [defaultEmbeddingDimension] - Default embedding dimension for collections created
- * or managed by this provider instance if not specified per collection.
- * @property {'cosine' | 'euclidean' | 'dotproduct'} [similarityMetric='cosine'] - Default similarity metric for new Pinecone indexes
- * created through this configuration, if applicable.
+ * This re-exports the runtime adapter config so the shared configuration contract
+ * stays aligned with the real Pinecone implementation and docs:
+ * - `apiKey`
+ * - `indexHost`
+ * - optional `namespace`
+ * - optional `defaultDimension`
+ * - optional `apiVersion`
  */
-export interface PineconeVectorStoreConfig extends VectorStoreProviderConfig {
-  type: 'pinecone';
-  apiKey: string; // Should be sourced from environment variables
-  environment: string;
-  defaultEmbeddingDimension?: number;
-  similarityMetric?: 'cosine' | 'euclidean' | 'dotproduct';
-}
+export type PineconeVectorStoreConfig = AdapterPineconeVectorStoreConfig;
 
 /**
  * Specific configuration for a Weaviate vector store.
@@ -255,16 +248,16 @@ export interface RagDataLayerConfig {
 const systemRagDataLayerConfig: RagDataLayerConfig = {
   vectorStoreManager: {
     managerId: 'main-vsm',
-    defaultProviderId: 'pinecone_main_prod',
+    defaultProviderId: 'qdrant_main_prod',
     defaultEmbeddingDimension: 1536, // System-wide default if not specified lower
     providers: [
       {
-        id: 'pinecone_main_prod', // Matches VectorStoreProviderConfig.id
-        type: 'pinecone',
-        apiKey: process.env.PINECONE_API_KEY!,
-        environment: process.env.PINECONE_ENVIRONMENT!,
+        id: 'qdrant_main_prod', // Matches VectorStoreProviderConfig.id
+        type: 'qdrant',
+        url: process.env.QDRANT_URL!,
+        apiKey: process.env.QDRANT_API_KEY,
         defaultEmbeddingDimension: 1536, // Provider-level default
-      } as PineconeVectorStoreConfig,
+      } as QdrantVectorStoreConfig,
       {
         id: 'in_memory_dev_store',
         type: 'in_memory',
@@ -277,8 +270,8 @@ const systemRagDataLayerConfig: RagDataLayerConfig = {
       dataSourceId: 'global_company_wiki',
       displayName: 'Global Company Wiki',
       description: 'Shared knowledge base for all company agents and employees.',
-      vectorStoreProviderId: 'pinecone_main_prod',
-      actualNameInProvider: 'company-wiki-prod-v2', // e.g., Pinecone index name
+      vectorStoreProviderId: 'qdrant_main_prod',
+      actualNameInProvider: 'company-wiki-prod-v2', // e.g., Qdrant collection name
       embeddingDimension: 1536, // Specific to this data source's content
       isDefaultQuerySource: true,
       metadataSchema: { "department": "string", "lastReviewed": "date" }
@@ -287,7 +280,7 @@ const systemRagDataLayerConfig: RagDataLayerConfig = {
       dataSourceId: 'user_personal_notes_main',
       displayName: 'User Personal Notes (Encrypted)',
       description: 'Personal notes and memories for individual users. Data segregated by user ID in metadata filters.',
-      vectorStoreProviderId: 'pinecone_main_prod',
+      vectorStoreProviderId: 'qdrant_main_prod',
       actualNameInProvider: 'user-notes-prod-encrypted',
       embeddingDimension: 1536,
       // This collection would typically be queried with a strong `userId` metadata filter.

src/core/vector-store/IVectorStore.ts

@@ -60,10 +60,10 @@ export type MetadataValue = MetadataScalarValue | MetadataScalarValue[];
 export interface MetadataFieldCondition {
   $eq?: MetadataScalarValue;        // Equal to
   $ne?: MetadataScalarValue;        // Not equal to
-  $gt?: number;                     // Greater than
-  $gte?: number;                    // Greater than or equal to
-  $lt?: number;                     // Less than
-  $lte?: number;                    // Less than or equal to
+  $gt?: number | string;            // Greater than. Strings are compared lexicographically (e.g. ISO timestamps).
+  $gte?: number | string;           // Greater than or equal to. Strings are compared lexicographically.
+  $lt?: number | string;            // Less than. Strings are compared lexicographically (e.g. ISO timestamps).
+  $lte?: number | string;           // Less than or equal to. Strings are compared lexicographically.
   $in?: MetadataScalarValue[];      // Value is one of these (for scalar fields)
   $nin?: MetadataScalarValue[];     // Value is not one of these (for scalar fields)
   $exists?: boolean;                // Field exists (true) or does not exist (false)
@@ -160,6 +160,28 @@ export interface QueryResult {
   stats?: Record<string, any>;
 }
 
+/**
+ * Options for enumerating documents by metadata without a query embedding.
+ */
+export interface MetadataScanOptions {
+  filter?: MetadataFilter;
+  limit?: number;
+  includeEmbedding?: boolean;
+  includeMetadata?: boolean;
+  includeTextContent?: boolean;
+  cursor?: string;
+  customParams?: Record<string, any>;
+}
+
+/**
+ * Result of a metadata-only document scan.
+ */
+export interface MetadataScanResult {
+  documents: RetrievedVectorDocument[];
+  nextCursor?: string;
+  stats?: Record<string, any>;
+}
+
 /**
  * Options for a vector store upsert (insert or update) operation.
  *
@@ -291,6 +313,15 @@ export interface IVectorStore {
     options?: QueryOptions,
   ): Promise<QueryResult>;
 
+  /**
+   * Optional: Enumerates documents using metadata-only filtering without
+   * requiring a query embedding.
+   */
+  scanByMetadata?(
+    collectionName: string,
+    options?: MetadataScanOptions,
+  ): Promise<MetadataScanResult>;
+
   /**
    * Optional: Hybrid retrieval combining dense vector similarity with lexical search.
    *