fix: Gemini api type mapping + underscore rendering in code

- model-presets: google-gemini → google-generative-ai (fixes #93, matches upstream OpenClaw config schema)
- markdown: underscore-to-italic regex now requires word boundary (fixes #92, my_variable no longer becomes italic)

src/lib/markdown.js

@@ -147,7 +147,7 @@ function inlineFormat(text) {
     .replace(/\*\*(.+?)\*\*/g, '<strong>$1</strong>')
     .replace(/\*(.+?)\*/g, '<em>$1</em>')
     .replace(/__(.+?)__/g, '<strong>$1</strong>')
-    .replace(/_(.+?)_/g, '<em>$1</em>')
+    .replace(/(?<!\w)_(.+?)_(?!\w)/g, '<em>$1</em>')
     .replace(/!\[([^\]]*)\]\(([^)]+)\)/g, (_, alt, src) => {
       const safeSrc = resolveImageSrc(src.trim())
       return `<img src="${safeSrc}" alt="${alt}" class="msg-img" onerror="this.onerror=null;this.style.display='none';this.insertAdjacentHTML('afterend','<span style=\\'color:var(--text-tertiary);font-size:12px\\'>[图片无法加载: ${escapeHtml(src)}]</span>')" />`

src/lib/model-presets.js

@@ -8,7 +8,7 @@ export const API_TYPES = [
   { value: 'openai-completions', label: 'OpenAI 兼容 (最常用)' },
   { value: 'anthropic-messages', label: 'Anthropic 原生' },
   { value: 'openai-responses', label: 'OpenAI Responses' },
-  { value: 'google-gemini', label: 'Google Gemini' },
+  { value: 'google-generative-ai', label: 'Google Gemini' },
 ]
 
 // 服务商快捷预设
@@ -23,7 +23,7 @@ export const PROVIDER_PRESETS = [
   { key: 'openai', label: 'OpenAI 官方', baseUrl: 'https://api.openai.com/v1', api: 'openai-completions' },
   { key: 'anthropic', label: 'Anthropic 官方', baseUrl: 'https://api.anthropic.com', api: 'anthropic-messages' },
   { key: 'deepseek', label: 'DeepSeek', baseUrl: 'https://api.deepseek.com/v1', api: 'openai-completions' },
-  { key: 'google', label: 'Google Gemini', baseUrl: 'https://generativelanguage.googleapis.com/v1beta', api: 'google-gemini' },
+  { key: 'google', label: 'Google Gemini', baseUrl: 'https://generativelanguage.googleapis.com/v1beta', api: 'google-generative-ai' },
   { key: 'nvidia', label: 'NVIDIA NIM', baseUrl: 'https://integrate.api.nvidia.com/v1', api: 'openai-completions', desc: '英伟达推理平台，支持 Llama、Mistral 等模型' },
   { key: 'ollama', label: 'Ollama (本地)', baseUrl: 'http://127.0.0.1:11434/v1', api: 'openai-completions' },
 ]