[improvement][supersonic] add text-to-sql evaluation (#696)

* [improvement] llm supports all models

* [improvement] alias convert to SemanticParseInfo

* [improvement] support join

* [improvement] add evaluation.py

* [improvement] add text2sql_evalution.py

* [improvement] add text2sql_evalution.py

* [improvement] add evalution

* [improvement] add evalution

* [improvement] add evalution

---------

Co-authored-by: zuopengge <hwzuopengge@tencent.com>

evaluation/README_CN.md

@@ -0,0 +1,5 @@
+# 评测流程
+
+1、正常启动项目(必须包括LLM服务)
+2、将要评测问题放到evalution/data目录下，如：internet.txt；将要评测问题对应的SQL也放到evalution/data目录下，如：gold_example_dusql.txt。
+3、执行evalution.sh脚本，主要包括构建表数据、获取模型预测结果，执行对比逻辑。可以在命令行看到执行准确率，错误case会写到同目录的eval.json文件中。