fix(mssql): add column/table comment support and fix Chinese encoding issue

[yunshaochu]

Description

Fix two issues with the MSSQL connector (MSSQLConnector):

1. Column and table comments cannot be read from MSSQL

MSSQL stores comments as extended properties (MS_Description in sys.extended_properties), but the original get_columns() method only queried INFORMATION_SCHEMA.COLUMNS which does not include them. This caused DB-GPT's NL2SQL to miss important schema context (e.g., column descriptions like "客户编号", "订单金额"), reducing SQL generation accuracy for Chinese-named databases.

Changes:

• get_columns(): Added LEFT JOIN sys.extended_properties to read column comments (MS_Description)
• New get_table_comment(): Reads table-level comments from sys.extended_properties
• table_simple_info(): Rewritten to reuse get_table_names()/get_columns(), now includes column comments and table comments in output
• Replaced repetitive .decode("utf-8") if isinstance(x, bytes) else x with _decode_if_bytes() helper

2. Chinese characters displayed as garbled text (mojibake)

When MSSQL server uses GBK encoding, Chinese characters are returned as Latin-1 encoded GBK strings, causing garbled output in query results and schema information.

Fix: Added _transform_val() that transcodes latin-1 → gbk, and overrode _query() and query_ex() to apply the transformation to all query results.

How Has This Been Tested?

1. Connected to MSSQL 2022 Docker container with Chinese-named tables (e.g., 我是大家庭, 超级变变变, 我是大聪明)
2. Verified get_columns() now returns column comments from MS_Description extended properties
3. Verified get_table_comment() correctly reads table-level comments
4. Verified table_simple_info() output includes comments: dbo.A1(B1(客户编号),B2(客户姓名),...) COMMENT[客户信息表]
5. Verified Chinese query results are no longer garbled after _transform_val() transcoding
6. Verified NL2SQL can now use column/table comments for better SQL generation
7. Ran make fmt, make fmt-check — all passed

Snapshots:

Before fix - get_columns() output (no comments, garbled):

{"name": "B1", "type": "int", "comment": None}  # Missing comment
# Chinese query results: ¿¿¿ (garbled)

After fix - get_columns() output (with comments, correct encoding):

{"name": "B1", "type": "int", "comment": "客户编号"}  # Comment read correctly
# Chinese query results: 客户编号 (correct)

table_simple_info() output:

Before: dbo.A1(B1,B2,B3,B4,B5,B6);
After:  dbo.A1(B1(客户编号),B2(客户姓名),B3(性别（男/女）),B4(所在城市),B5(注册日期),B6(账户余额)) COMMENT[客户信息表];

Checklist:

• My code follows the style guidelines of this project
• I have already rebased the commits and make the commit message conform to the project standard.
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation (N/A - bug fix for existing connector)
• Any dependent changes have been merged and published in downstream modules (N/A)

[Aries-ckt]

LGTM.Thanks for your contribution.

[chenliang15405]

LGTM

[chenliang15405]

@yunshaochu Thanks for your contribution, please fix code quality with command make fmt
And Maybe the encoding conversion between GBK and latin-1 may fail in non-Chinese environments. Please ensure that exception handling covers all scenarios.