chore(repo): clarify autonomous research network and remove unused code

[echobt]

Summary

• Present Cortex as an autonomous research network, with an investor-readable overview and audience-based documentation index instead of implementation-language-first positioning.
• Ground the explanation in all six pages of the whitepaper, especially §7: reproducible findings and reusable recipes can contribute to a shared research collection, rather than only selecting an isolated winning checkpoint. Explain that the proposed synthesis/adoption agent is not implemented and that complementary findings still need combined evaluation.
• Add a paper-to-code comparison. Document existing limitations accurately: partial Python judging and recipe reproduction, in-memory Proof records, epoch-zero HTTP state without automatic reward emission, and differences between proposed scoring equations and current payout functions. This PR does not implement those mechanisms.
• Remove unconsumed chain stubs and their obsolete feature, unused private helpers, the superseded similarity-v2 prompt, the unused Proof binary-average helper, unused dependencies, and a redundant dead-code allowance. Preserve ChainClient, FakeChain, production chain-live, active v3 prompts, and Proof's consumed prism-* harvest stack.
• Share metadata storage projection across v14/v15, add map/plain-entry regressions, and test the gateway's actual URL helper. Correct stale Bounty allocation, Proof aggregation, Compose topology/profile, persistence, and retired-challenge guidance.

See docs/CLEANUP.md for deletion/retention evidence and docs/WHITEPAPER.md for the proposal-versus-implementation comparison. The PDF itself is unchanged.

Greptile

• Greptile has reviewed this PR; findings are fixed or answered
• If the bot was silent, I commented @greptileai review

Three local simplification reviews covered the entire 50-file change: reuse, quality/compatibility provenance, and efficiency. No introduced actionable code issues were found. Their routing-persistence documentation clarification was applied.

Test plan

• cargo fmt --all -- --check
• cargo clippy --workspace --all-targets -- -D warnings
• cargo test --workspace: 1,119 passed, 0 failed, 11 ignored
• cargo-deny check using an isolated installation: advisories, bans, licenses, and sources pass; non-blocking dependency warnings remain
• cargo run -p xtask -- loc-cap
• cargo run -p xtask -- consensus-lint
• cargo run -p xtask -- spec-check
• cargo run -p xtask -- design-check
• cargo run -p xtask -- external-docs-check
• cargo clippy -p validator-bin --features dcap --all-targets -- -D warnings
• bash deploy/scripts/assert-compose-matrix.sh
• Python evaluation contract tests: 12 passed with isolated test dependencies
• Repository pre-commit compile gate, diff whitespace checks, and 225 relative Markdown file links

Not run: paid Lium/GPU experiments, deployment, live on-chain submission, or the 11 ignored Rust tests. Local HTTP/challenge and signed-leaf integration tests are included in the workspace suite; these are not presented as proof of live scientific reproduction or payments.

Risk

No change to scoring formulas, emission shares, trust roots, signature domains, applied migrations, deployment settings, or pinned image digests. Compose changes are comments only. Removed Rust surfaces had no workspace consumers; the chain crate is unpublished. Existing scientific/economic readiness gaps are documented rather than hidden or altered during cleanup.

Naming

BASE_*, /opt/base, /run/base, existing registry paths, cryptographic domain tags, compatibility routes, frozen specifications, historical miner links, and operational evidence remain intact.

[echobt]

@greptileai review

[greptile-apps]

Greptile Summary

Summary

• Reframes Cortex as an autonomous research network while documenting the boundary between the Proof proposal and current implementation.
• Aligns deployment and operational guidance with the master, validator, gateway, challenge, and optional updater topology.
• Removes obsolete Rust code and unused dependency declarations without changing the exercised gateway forwarding, metadata snapshot, or Proof allocation behavior.

Merge safety

Safe to merge. No issues requiring changes were identified.

Confidence Score: 5/5

Safe to merge: no actionable defects were identified in the reviewed changes.

No final findings remain after exercising gateway request forwarding, live metadata snapshot projection, and Proof scoring and emission behavior before and after the changes.

Files Needing Attention: None.

T-Rex Logs

What T-Rex did

• Authored and ran a focused gateway integration test against both revisions, exercising root and nested challenge routes with query strings, path variations, and a POST flow through a live upstream.
• Ran the focused metadata snapshot tests and live Finney metadata snapshot against both revisions, confirming map hasher/default retention, plain storage handling, and live storage probing.
• Ran the focused Proof scoring, challenge, evaluation, and HTTP test suite against both revisions; all 40 tests passed, including payout allocation and signed-emission coverage, and dependent crates compiled after removing mean_lattice with no remaining scoped production call sites.
• Executed exact authored validation using gateway-proxy-refactor-validate.rs and the accompanying command script gateway-proxy-refactor-command.sh; the process copied the test into an isolated HEAD^ worktree and the current checkout, ran cargo test -p gateway --test gateway_proxy_refactor_validate -- --nocapture, and collected the full command outputs.
• Ran the exact metadata snapshot projection verify script; the pre-change capture proves baseline behavior and the PR capture confirms map hasher/default retention, plain entry handling, and live storage probing after the refactor, with identical pre/post lockfile SHA-256 checksums.

_{Ran code and verified through T-Rex}

_{Reviews (1): Last reviewed commit: "chore(repo): clarify research network an..." | Re-trigger Greptile}