springboot + shardingSphere5.0 + Dubbo2.7.x实现分库分表的情况下，集成 + seata1.4.2后，报出以下异常： Could not found any index in the table: xxx

各位大神好，我在springboot + shardingSphere5.0 + Dubbo2.7.x实现分库分表(3个库，t_order分4张表)的情况下，集成 + seata1.4.2后，报出以下异常：
Could not found any index in the table: xxx，查看了下源码，在io.seata.rm.datasource.sql.struct.cache.MysqlTableMetaCache#resultSetMetaToSchema方法中，会判断所有index是否为空，

```
try (ResultSet rsColumns = dbmd.getColumns(catalogName, schemaName, tableName, "%");
ResultSet rsIndex = dbmd.getIndexInfo(catalogName, schemaName, tableName, false, true)) {
while (rsIndex.next()) {
String indexName = rsIndex.getString("INDEX_NAME");
String colName = rsIndex.getString("COLUMN_NAME");
ColumnMeta col = tm.getAllColumns().get(colName);

if (tm.getAllIndexes().containsKey(indexName)) {
IndexMeta index = tm.getAllIndexes().get(indexName);
index.getValues().add(col);
} else {
IndexMeta index = new IndexMeta();
index.setIndexName(indexName);
index.setNonUnique(rsIndex.getBoolean("NON_UNIQUE"));
index.setIndexQualifier(rsIndex.getString("INDEX_QUALIFIER"));
index.setIndexName(rsIndex.getString("INDEX_NAME"));
index.setType(rsIndex.getShort("TYPE"));
index.setOrdinalPosition(rsIndex.getShort("ORDINAL_POSITION"));
index.setAscOrDesc(rsIndex.getString("ASC_OR_DESC"));
index.setCardinality(rsIndex.getInt("CARDINALITY"));
index.getValues().add(col);
if ("PRIMARY".equalsIgnoreCase(indexName)) {
index.setIndextype(IndexType.PRIMARY);
} else if (!index.isNonUnique()) {
index.setIndextype(IndexType.UNIQUE);
} else {
index.setIndextype(IndexType.NORMAL);
}
tm.getAllIndexes().put(indexName, index);

}
}
if (tm.getAllIndexes().isEmpty()) {
throw new ShouldNeverHappenException("Could not found any index in the table: " + tableName);
}
}
return tm;
```

【(tm.getAllIndexes().isEmpty()】这行条件成立就会抛出异常，请问此处的判断是什么用意，谢谢。
PS：表中都有主键

Start by reading io.seata.rm.datasource.sql.struct.cache.MysqlTableMetaCache#resultSetMetaToSchema and the shown getIndexInfo handling. Reproduce the Spring Boot, ShardingSphere, Dubbo, and Seata setup against the described three databases and four t_order tables, then inspect the metadata returned for the primary keys. Done means explaining why no indexes are found and documenting or resolving the resulting exception.

Written by the indexing model from the issue text.