feat: add AI_EMBEDDING_SUPPORTS_MRL env var to decouple pgvector dimensions from model API (#2079)

Some embedding models don't accept a `dimensions` parameter. This adds
an optional env var that controls whether the dimension is sent to the
model API, while always using it for pgvector indexing. Preset models
have this handled automatically; the env var allows explicit override
for custom models.

apps/server/src/integrations/environment/environment.service.ts

@@ -259,6 +259,12 @@ export class EnvironmentService {
     );
   }
 
+  getAiEmbeddingSupportsMrl(): boolean | undefined {
+    const val = this.configService.get<string>('AI_EMBEDDING_SUPPORTS_MRL');
+    if (val === undefined || val === null || val === '') return undefined;
+    return val === 'true';
+  }
+
   getOpenAiApiKey(): string {
     return this.configService.get<string>('OPENAI_API_KEY');
   }

apps/server/src/integrations/environment/environment.validation.ts

@@ -117,6 +117,12 @@ export class EnvironmentVariables {
   @IsString()
   AI_EMBEDDING_DIMENSION: string;
 
+  @IsOptional()
+  @ValidateIf((obj) => obj.AI_EMBEDDING_SUPPORTS_MRL)
+  @IsIn(['true', 'false'])
+  @IsString()
+  AI_EMBEDDING_SUPPORTS_MRL: string;
+
   @ValidateIf((obj) => obj.AI_DRIVER)
   @IsString()
   @IsNotEmpty()