Interface ContentPipeline

Contract that every article-type plugin must implement.

Lifecycle:

1. fetch – Retrieve raw data from MCP tools (may partially fail).
2. transform – Convert raw data into article payloads per language.
3. generate – Render HTML via generateArticleHTML.
4. validate – Verify HTML structure before writing.
5. write – Persist files to news/ directory.

Implementations are responsible for graceful degradation: if the fetch stage partially fails they should log a warning and continue with whatever data is available rather than throwing.