docs(blogs): drop dead scws download links

Both URLs in the tip were dead. The GitHub one referenced an account that
does not exist, and the Gitea one is retired.

Point at the upstream SCWS project instead, and stop promising a
pre-built Windows binary — upstream publishes source only, and the 1.2.3
the dead link referenced was OnixByte's own build.

docs/en-gb/blogs/postgresql-zhparser-fuzzy-search.md

@@ -73,7 +73,7 @@ This step requires `pg_config` from the self-compiled PostgreSQL installation.
    make USE_PGXS=1 PG_CONFIG=/usr/local/pgsql/bin/pg_config install
    ```
 
-   *Note: The `install` step automatically places `zhparser.so` into PG's `pkglibdir` and extension scripts into the `extension` directory.*
+   _Note: The `install` step automatically places `zhparser.so` into PG's `pkglibdir` and extension scripts into the `extension` directory._
 
 ### Resolving Dynamic Library Dependencies
 
@@ -152,8 +152,15 @@ Leverage the structural parsing capabilities of **`companynameparser`** to strip
 
 ### Lexicon Compilation and Deployment
 
+<!-- TODO(link): this tip used to offer OnixByte's pre-compiled Windows build of
+scws. Both URLs were dead — the GitHub one referenced an account that does not
+exist, and the Gitea one is being retired. Pointing at upstream in the meantime;
+restore a download link here if one is published again. -->
+
 :::tip
-Users compiling `xdb` binary dictionary files on Windows can visit OnixByte’s [GitHub](https://github.com/onixbyte/scws/releases/tag/1.2.3) or [GitLab](https://git.onixbyte.cn/onixbyte/scws/-/releases/1.2.3) pages to download the native scws command-line tool for Windows, pre-compiled using MingW.
+The native scws command-line tool for Windows is compiled with MingW. Source and
+build instructions are available from the upstream
+[SCWS project](https://github.com/hightman/scws).
 :::
 
 Convert the text dictionary to SCWS's efficient binary format (XDB):
@@ -208,5 +215,6 @@ SELECT * FROM ts_debug('chinese', '元一能源');
 ```
 
 **Optimisation Notes:**
+
 - **Explicit Weight Compensation**: This is the key technique that resolved the "元一" tokenisation failure (shown as `x`).
 - **Distinguish Restart from Reload**: `shared_preload_libraries` must be activated via `restart`, not a simple reload.

docs/zh-hans/blogs/postgresql-zhparser-fuzzy-search.md

@@ -73,7 +73,7 @@ SCWS 是 `zhparser` 依赖的底层核心，必须先行安装。
    make USE_PGXS=1 PG_CONFIG=/usr/local/pgsql/bin/pg_config install
    ```
 
-   *注：`install` 步骤会自动将 `zhparser.so` 放入 PG 的 `pkglibdir`，将脚本放入 `extension` 目录。*
+   _注：`install` 步骤会自动将 `zhparser.so` 放入 PG 的 `pkglibdir`，将脚本放入 `extension` 目录。_
 
 ### 解决动态链接库依赖
 
@@ -152,8 +152,13 @@ ALTER TEXT SEARCH CONFIGURATION chinese ADD MAPPING FOR n,v,a,i,e,l,t,b WITH sim
 
 ### 词库编译与部署
 
+<!-- TODO(link): 该提示原本指向 OnixByte 预编译的 Windows 版 scws，但两个链接
+都是死链 —— GitHub 那个引用的账号并不存在，Gitea 那个正在下线。暂时指向上游；
+若日后重新发布下载地址，在此恢复。 -->
+
 :::tip
-对于使用 Windows 进行编译 `xdb` 二进制词典文件的用户，可以前往 OnixByte 的 [GitHub](https://github.com/onixbyte/scws/releases/tag/1.2.3) 或 [GitLab](https://git.onixbyte.cn/onixbyte/scws/-/releases/1.2.3) 页面下载使用 MingW 预编译好的，适用于 Windows 的原生 scws 命令行工具。
+适用于 Windows 的原生 scws 命令行工具使用 MingW 编译。源码与编译说明见上游
+[SCWS 项目](https://github.com/hightman/scws)。
 :::
 
 将文本词典转换为 SCWS 高效二进制格式（XDB）：
@@ -208,5 +213,6 @@ SELECT * FROM ts_debug('chinese', '元一能源');
 ```
 
 **优化建议：**
+
 - **显式提及权重补偿**：这是解决"元一"分词失败（显示为 `x`）的关键技术点。
 - **区分重启与重载**：明确 `shared_preload_libraries` 必须通过 `restart` 激活。