fix: field normalization, AI connection validation, autofix filter (#581) (#638)

- Normalize name→nodeName and id→nodeId for node-targeting operations in
  the Zod schema transform, so LLMs using natural field names no longer
  get "Node not found" errors
- Replace hardcoded ALL_CONNECTION_TYPES with dynamic iteration so AI
  sub-nodes (ai_outputParser, ai_document, ai_textSplitter, etc.) are
  not flagged as disconnected during save
- Add .catchall() to workflowConnectionSchema and extend connection
  reference validation to cover all connection types, not just main
- Fix filterOperationsByFixes ID-vs-name mismatch: typeversion-upgrade
  operations now include nodeName alongside nodeId, and the filter checks
  both fields

Conceived by Romuald Członkowski - https://www.aiadvisors.pl/en

Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

dist/types/n8n-api.d.ts

@@ -259,6 +259,7 @@ export interface WebhookRequest {
 }
 export interface McpToolResponse {
     success: boolean;
+    saved?: boolean;
     data?: unknown;
     error?: string;
     message?: string;
@@ -266,6 +267,7 @@ export interface McpToolResponse {
     details?: Record<string, unknown>;
     executionId?: string;
     workflowId?: string;
+    operationsApplied?: number;
 }
 export type ExecutionMode = 'preview' | 'summary' | 'filtered' | 'full' | 'error';
 export interface ExecutionPreview {