PACSeries v0.3: packaging, consistency pass, and method framing

[lornecodes]

Prepares PACSeries v0.3 (Papers 7–12) for release. Scope locked to the new papers; no changes to the published v0.1/v0.2.

What's in it

• Content & consistency — fixed cross-ref numbering, misattributed experiment IDs, score labels, a self-citation, DOI labels (concept vs v0.2 record), and metadata; added Paper 10's previously-omitted Type-A verification numbers.
• Voice — de-LLM-ism pass across all six papers (no claims/numbers/structure changed).
• Packaging — full reproducibility packages to the v0.2 gold standard: Code/experiments + reproduce.py + trace.yaml + requirements.txt + generate_figures.py, Data/results, Figures/, meta.yaml, README.md, paper.tex. Repo-relative model (reproduce.py injects the milestone core/ onto PYTHONPATH). Papers 7–11 run-verified; Paper 12 is data-gated (external SDSS/XQR catalogs).
• Corrections (Imperfection-Engine collapse events) — adversarial verification of every headline number against its result JSON caught six over-attributions, all corrected: Paper 11 Bell "Tsirelson saturation" (graph-derived rotation reaches 2.514, not 2.828); Paper 8 bounce time (1.0 t_P, not the 0.72 cascade tick), Hawking CV, Page remnant; Paper 10 ds² precision + exp_16 disclosure; Paper 7 σ/m.
• Method framing — new PACSeries/HOW_TO_READ.md (methodological statement of the evolving-program / A-B-C-tiered / reproducible stance) + a consistent "Epistemic status" block in every paper.
• New tooling: md_to_tex.py (pandoc-free md→LaTeX).

See .changelog/20260706_115131_pacseries-v0.3-packaging.md for the detailed log.

Notes for review

• Paper 12 experiments need external catalogs; packaged but not executed here.
• Two scores kept at their state-doc-consistent values with the gaps disclosed (P10 M13 51/60 + exp_16 noted; P7 M6 34/40 + drift note).
• Milestone R (Paper 12's exp_24 reference) is not in foundational/experiments.

🤖 Generated with Claude Code