Support Ollama use: chat|memory and parallel embed beside VL.

Pull nomic-embed-text for Assistent memory, write ollama-roles.json, CPU Modelfile, and raise MAX_LOADED_MODELS/NUM_PARALLEL to 2.

Co-authored-by: Cursor <cursoragent@cursor.com>

ollama-models.example.yaml

@@ -1,12 +1,18 @@
 # Copy to ollama-models.yaml (gitignored). Used when LLM_RUNTIME=ollama.
 # name = exact tag for `ollama pull`.
-# Requirement: uncensored (abliterated) + Russian. Empty models: [] → no pull.
+# use: chat — Assistent header select; use: memory — settings memory model.
+# Requirement: uncensored (abliterated) + Russian for chat. Empty models: [] → no pull.
 
 models:
   # Recommended (~6GB): vision + RU/EN, abliterated
   - name: huihui_ai/qwen2.5-vl-abliterated:7b
+    use: chat
     default: true
 
+  # Vector memory (CPU via num_gpu 0 Modelfile after pull) — not shown in chat select
+  - name: nomic-embed-text
+    use: memory
+
   # Presets (setup / up --ollama-preset …):
   # light — huihui_ai/qwen2.5-vl-abliterated:3b   (~3GB)
   # text  — huihui_ai/qwen2.5-abliterate:7b       (~5GB, no vision)