Implement graphify Docker graph-report smoke

[yvette-carlisle]

Summary

• add a Docker-only graphify graph/report smoke command that installs graphify in the baseline runner and records typed setup/run/result artifacts
• wire graphify into the live adapter sweep and generated real_world_job fixture path
• update the external adapter manifest/tests and research docs so graphify is an implementation-backed research gate without broad quality claims

Verification

• python3 -m py_compile scripts/graphify-docker-graph-report-smoke.py
• jq empty apps/elf-eval/fixtures/real_world_external_adapters/memory_projects_manifest.json
• cargo make graphify-docker-graph-report-smoke
• cargo test -p elf-eval --test real_world_job_benchmark real_world_report_includes_external_adapter_coverage_manifest -- --nocapture
• cargo make fmt
• cargo make lint-fix
• cargo make checks
• git diff --check
• semantic drift audit: pass; graphify docs now distinguish checked-in research_gate from generated live artifact

Decodex note

XY-889 attempt 1 stalled with process_stopped/active_lease=false while retaining useful work. I inspected Decodex evidence, continued from the retained worktree, and did not commit Decodex runtime control files.