docs(config): add .gemini-refiner.example.json for local Ollama configuration

Documents the semantic block fields for wiring LocalOpenAiProvider to Ollama
(port 11434) or LM Studio. Active config goes in .gemini-refiner.json which
is gitignored. Example shows gemma3 models with 120s timeout for local use.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: Aitomates

universal-refiner/.gemini-refiner.example.json

@@ -0,0 +1,13 @@
+{
+  "_comment_usage": "Copy this file to .gemini-refiner.json and fill in your values. This file is safe to commit.",
+  "_comment_fields": "All fields are optional. Omit a section to use built-in defaults.",
+  "semantic": {
+    "_comment": "Configure the local OpenAI-compatible provider (Ollama, LM Studio, etc.)",
+    "localEnabled": true,
+    "baseUrl": "http://localhost:11434/v1",
+    "models": ["gemma3:12b", "gemma3"],
+    "mcpSamplingEnabled": true,
+    "timeoutMs": 120000,
+    "temperature": 0.2
+  }
+}