refactor(config): extract hardcoded model defaults to constants

[Copilot]

Hardcoded model default values scattered across config classes make synchronization with env.example error-prone.

Changes

• Extract model defaults to module-level constants with explicit mapping to env vars:
  • DEFAULT_LLM_MODEL → LLM_MODEL
  • DEFAULT_EMBEDDING_MODEL → EMBEDDING_MODEL
  • DEFAULT_RERANK_MODEL → RERANK_MODEL
  • DEFAULT_DS_OCR_MODEL → DS_OCR_MODEL

# ==================== Default Model Constants ====================
# These default values must stay in sync with env.example
# They are only used when the corresponding environment variable is not set:
#   - DEFAULT_LLM_MODEL       -> LLM_MODEL
#   - DEFAULT_EMBEDDING_MODEL -> EMBEDDING_MODEL
#   - DEFAULT_RERANK_MODEL    -> RERANK_MODEL
#   - DEFAULT_DS_OCR_MODEL    -> DS_OCR_MODEL

DEFAULT_LLM_MODEL = "seed-1-6-250615"
DEFAULT_EMBEDDING_MODEL = "Qwen/Qwen3-Embedding-0.6B"
DEFAULT_RERANK_MODEL = "Qwen/Qwen3-Reranker-8B"
DEFAULT_DS_OCR_MODEL = "deepseek-ai/DeepSeek-OCR"

Behavior unchanged—pydantic-settings still reads from env vars first, falling back to these constants only when unset.

Original prompt

---

This section details on the original issue you should resolve

<issue_title>Rerank model default hardcoded - should use env variable</issue_title>
<issue_description>## 问题描述
Rerank 模型默认值硬编码。

受影响的文件

• src/config.py 行 75

硬编码值

• default="Qwen/Qwen3-Reranker-8B"

问题

虽然有 RERANK_MODEL 环境变量，但代码中也硬编码。

解决方案

应该移除硬编码的默认值或保持一致。</issue_description>

Comments on the Issue (you are @copilot in this section)

• Fixes Rerank model default hardcoded - should use env variable #19

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.