fix: 修复中介导入成功条数计算错误

问题：
- 导入成功条数显示为负数
- 原因：成功数量计算使用 validRecords.size() - failures.size()
- 但没有使用实际的数据库操作返回值

修复：
- saveBatchWithUpsert 和 saveBatch 方法现在返回 int
- 累加实际的数据库影响行数
- 使用 actualSuccessCount 变量跟踪真实成功数量

影响范围：
- CcdiIntermediaryPersonImportServiceImpl
- CcdiIntermediaryEntityImportServiceImpl

doc/test-data/purchase_transaction/node_modules/xmlchars/README.md

@@ -0,0 +1,33 @@
+Utilities for determining whether characters belong to character classes defined
+by the XML specs.
+
+## Organization
+
+It used to be that the library was contained in a single file and you could just
+import/require/what-have-you the `xmlchars` module. However, that setup did not
+work well for people who cared about code optimization. Importing `xmlchars`
+meant importing *all* of the library and because of the way the code was
+generated there was no way to shake the resulting code tree.
+
+Different modules cover different standards. At the time this documentation was
+last updated, we had:
+
+* `xmlchars/xml/1.0/ed5` which covers XML 1.0 edition 5.
+* `xmlchars/xml/1.0/ed4` which covers XML 1.0 edition 4.
+* `xmlchars/xml/1.1/ed2` which covers XML 1.0 edition 2.
+* `xmlchars/xmlns/1.0/ed3` which covers XML Namespaces 1.0 edition 3.
+
+## Features
+
+The "things" each module contains can be categorized as follows:
+
+1. "Fragments": these are parts and pieces of regular expressions that
+correspond to the productions defined in the standard that the module
+covers. You'd use these to *build regular expressions*.
+
+2. Regular expressions that correspond to the productions defined in the
+standard that the module covers.
+
+3. Lists: these are arrays of characters that correspond to the productions.
+
+4. Functions that test code points to verify whether they fit a production.