docs(research): map selected bootstrap proof obligations

[petalas]

Summary

bottleneck brief
  -> selected bootstrap obligation map
       retained cutoff/cap audits: #1956, #1965, #1990
       conditional BIT stock:     #2084, #2120
       conditional uniform time:  #2087
       fresh-field averaging:     #2097
       implementation and later-stage obligations remain

Add docs/research/selected-bootstrap-scope-map.md and one onboarding link. The map separates reviewed retained scopes from four pinned open proposals, ordinary BIT m=64 from complex m=73, sufficient coefficient caps from derived conditional coefficients, and time H2 from stock H3. It includes exact replay heads, commands, attribution and open next questions. No mathematical result is claimed as new.

Task scope-navigation, consolidation, priority revision 4; nonce 9540ac7c50e97c48ffce51501dabc841. Scope refined to this proof-dependency map. Base and refreshed canonical main are 0fdd7d82de03e43c9e5a1000dfbcd380d1b687d2. Canonical claim. PR #2133 already covers the P104 alternative and is credited rather than duplicated.

Evidence

• Before: the bottleneck brief links the retained unknown-stock audit but provides no route through the newer conditional BIT coefficient, uniform-time and fresh-fiber results.
• After: the new map records each result's premises and exclusions, with exact pinned public source links. All seven registered replays passed on Linux x86_64 / CPython 3.12.15, including their declared controls. Four open-head replays used separate clean sparse worktrees:

PR	Exact head	Package under research/
2084	8001a23b003bec5dff5e61edec907b652c475932	selected-bit-parent-reserve
2120	378593574261e0b0915e7e8a83147aa640274fc8	selected-bit-ceiling-free-stock
2087	b5ffc48e3dd5c92f591bdb4fabafa2c8e8b33970	selected-bootstrap-uniform-time
2097	1d3ee0926656d5260c6cb1c0b81964ffe9fe1377	selected-cover-fresh-fibers

Run python3.12 -Werror -B scripts/research_package.py --package research/PACKAGE at each cited head. The three retained replays are bootstrap-cutoff-carry, bootstrap-cutoff-time-space-split and selected-bootstrap-stock-audit on the stated main.

Also passed:

python3.12 -Werror -B -m unittest discover -s tests -p 'test_research_*.py'
python3 -B scripts/research_baseline.py
git diff --cached --check

All 65 workflow tests pass. The document's exact arithmetic snippet passes; changing its 6748 equality to 6749 fails. Eighteen relative file links and all four pinned Git source paths resolve. The selected record retains SHA256 10b9e340fd8e354a37a0c4813c90e5de1dd42044b5f6aca4a3fc5a42b685fa32.

Pre-submission overlap covered 428 open PRs and 864 Discussion records with complete pagination at 22:35/22:36 UTC. The four cited open heads were rechecked unchanged. This is metadata coverage, not a novelty certificate.

Merge Danger

Door: two-way.

Blast Radius: documentation.

No selected record, priority allocation, frozen evidence or verifier is changed. Open-PR proofs remain proposals. Replaying their authors' checks is not an independently implemented mathematical audit or human review. The actual packet/compiler, completed-leaf, uniform-toll, hosting, precision/tape and later-stage obligations remain; no full selected cold replay, formal-kernel proof, hosted CI pass or new κ is claimed.

Mathematical credit remains with Josie Elliston, ToikY, Adam Gall, Nick Petalas, Sricharan Vagvala and the selected PR589/PR555, PR185/PR387, PR1094/PR1087 and cover/borrowing lineage documented in the map.