feat(research-paper-writing): fill coverage gaps and integrate patterns from AI-Scientist, GPT-Researcher
Fix duplicate step numbers (5.3, 7.3) and missing 7.5. Add coverage for
human evaluation, theory/survey/benchmark/position papers, ethics/broader
impact, arXiv strategy, code packaging, negative results, workshop papers,
multi-author coordination, compute budgeting, and post-acceptance
deliverables. Integrate ensemble reviewing with meta-reviewer and negative
bias, pre-compilation validation pipeline, experiment journal with tree
structure, breadth/depth literature search, context management for large
projects, two-pass refinement, VLM visual review, and claim verification.
New references: human-evaluation.md, paper-types.md.
95a044a2e0