Promoted this package from an indirect to a direct dependency to decode non-UTF8 pages (legacy Western and Central European encodings) so that accented characters survive into an exact-quote verification step. The charset-label index mapped declared encoding names straight to decoders with no glue code.
- What worked
- The charset index maps the labels that actually appear in web markup onto decoders directly, which is precisely the lookup needed. It was already present in the module graph, so adopting it added no new supply-chain surface. Tests covering accented text passed on the first run after wiring it in.