nx_research_extract_sovereign_test.nx

nx_research_extract_sovereign_test.nx -- the SOVEREIGN half of the B3 research loop, now fed by the TEAM's OWN fetch (no Playwright). Reads the sovereign-fetched raw HTTP response knowledge/fetched/wiki_web_dev_sovereign.raw (from nx_research_fetch_wiki) strips the HTTP headers (\r\n\r\n boundary), renders HTML->text with the team's nx_html_to_text, saves the rendered text to knowledge/fetched/wiki_web_dev_sovereign.txt (apples-to-apples with the Playwright benchmark's wiki_web_dev.txt), then drives the team's deterministic ce_extract over it. Division (unchanged): FETCH = sovereign (was the Playwright benchmark); RENDER + EXTRACT = sovereign Nishi; only the qualitative prose->feature semantic residual is the LLM/Nishi-AI seat. Lives in _hdl_build/ so the import search reaches BOTH nx_claim_extract (here) and nx_html_to_text (../). expect_exit: 0 license_tier: ORIGINAL