fix(drafting): Drafting-Engine auf prod reparieren — RAG via ai-sdk + OVH-LLM-Kaskade

Die Drafting-Engine (Dokument-Entwurf, v2-Pipeline, Validierung, Drafting-Chat,
Vendor-Vertragspruefung) war auf prod doppelt tot:
- RAG ueber bp-core-rag-service:8097 (existiert auf prod nicht)
- LLM ueber OLLAMA_URL/api/chat mit qwen2.5vl (prod = ollama-embed, kein Chat-Modell)

Fix (analog zum Compliance-Advisor):
- rag-query.ts -> ai-compliance-sdk /sdk/v1/rag/search (bge-m3, prod-erreichbar).
- Neue lib/sdk/drafting-engine/llm-cascade.ts: OVH/LiteLLM (gpt-oss-120b) zuerst,
  Ollama als Dev-Fallback; cascadeComplete (JSON) + cascadeStream. Das Backend nutzt
  OVH+JSON bereits erfolgreich auf prod (extract-datasheet).
- 5 Aufrufstellen (draft-helpers, draft-helpers-v2, validate, chat, vendor-review)
  auf die Kaskade umgestellt; keine direkten Ollama-Calls mehr.
- Tests: llm-cascade + rag-query aktualisiert.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

admin-compliance/app/api/sdk/drafting-engine/chat/route.ts

@@ -11,9 +11,7 @@ import { queryRAG } from '@/lib/sdk/drafting-engine/rag-query'
 import { DOCUMENT_RAG_CONFIG } from '@/lib/sdk/drafting-engine/rag-config'
 import { readSoulFile } from '@/lib/sdk/agents/soul-reader'
 import type { ScopeDocumentType } from '@/lib/sdk/compliance-scope-types'
-
-const OLLAMA_URL = process.env.OLLAMA_URL || 'http://host.docker.internal:11434'
-const LLM_MODEL = process.env.COMPLIANCE_LLM_MODEL || 'qwen2.5vl:32b'
+import { cascadeStream } from '@/lib/sdk/drafting-engine/llm-cascade'
 
 // Fallback SOUL prompt (used when .soul.md file is unavailable)
 const FALLBACK_DRAFTING_PROMPT = `# Drafting Agent - Compliance-Dokumententwurf
@@ -81,66 +79,20 @@ export async function POST(request: NextRequest) {
     ]
 
     // 4. Call LLM with streaming
-    const ollamaResponse = await fetch(`${OLLAMA_URL}/api/chat`, {
-      method: 'POST',
-      headers: { 'Content-Type': 'application/json' },
-      body: JSON.stringify({
-        model: LLM_MODEL,
-        messages,
-        stream: true,
-        think: false,
-        options: {
-          temperature: mode === 'draft' ? 0.2 : 0.3,
-          num_predict: mode === 'draft' ? 16384 : 8192,
-          num_ctx: 8192,
-        },
-      }),
-      signal: AbortSignal.timeout(120000),
+    // 4. LLM-Kaskade (OVH -> Ollama) -> Plain-Text-Stream
+    const stream = await cascadeStream(messages, {
+      temperature: mode === 'draft' ? 0.2 : 0.3,
+      maxTokens: mode === 'draft' ? 16384 : 8192,
+      timeoutMs: 120000,
     })
 
-    if (!ollamaResponse.ok) {
-      const errorText = await ollamaResponse.text()
-      console.error('LLM error:', ollamaResponse.status, errorText)
+    if (!stream) {
       return NextResponse.json(
-        { error: `LLM nicht erreichbar (Status ${ollamaResponse.status})` },
+        { error: 'LLM nicht erreichbar (weder OVH noch Ollama)' },
         { status: 502 }
       )
     }
 
-    // 5. Stream response back
-    const encoder = new TextEncoder()
-    const stream = new ReadableStream({
-      async start(controller) {
-        const reader = ollamaResponse.body!.getReader()
-        const decoder = new TextDecoder()
-
-        try {
-          while (true) {
-            const { done, value } = await reader.read()
-            if (done) break
-
-            const chunk = decoder.decode(value, { stream: true })
-            const lines = chunk.split('\n').filter((l) => l.trim())
-
-            for (const line of lines) {
-              try {
-                const json = JSON.parse(line)
-                if (json.message?.content) {
-                  controller.enqueue(encoder.encode(json.message.content))
-                }
-              } catch {
-                // Partial JSON, skip
-              }
-            }
-          }
-        } catch (error) {
-          console.error('Stream error:', error)
-        } finally {
-          controller.close()
-        }
-      },
-    })
-
     return new NextResponse(stream, {
       headers: {
         'Content-Type': 'text/plain; charset=utf-8',