FE-1253: Add prospect research regression

[kostandinang]

Why

Prospect research needs stateful full-stack regression coverage without turning every case into a rigorous end-to-end campaign.

What

Add an exploratory prospect mission and a frozen execution-only React, Node.js, TypeScript, and SQLite case. Its deterministic oracle covers browser, HTTP, SQLite, export, restart, provider-failure, and runtime-network behavior while reusing the shared packet and attempt contracts.

The case does not add a prospect-specific 2x2 study, requirement ledger, scored-provider gate, or comparative claim.

How to test

• npm run check
• npm run build
• npx vitest --run --maxWorkers=1 src/dev/execution-comparison/__tests__/prospect-research-workspace-oracle.test.ts

[kostandinang]

• FE-1253: Add prospect research regression #369 👈 (View in Graphite)
• FE-1241: Add isolated brownfield comparison cases #362 : 1 other dependent PR (#370 )
• next

This stack of pull requests is managed by Graphite. Learn more about stacking.

[cursor]

PR Summary

Medium Risk
Touches shared execution-comparison contracts, oracle dispatch, and Brunch seeding paths used by other cases; mistakes could break case registration or brownfield/greenfield parsing, though coverage is heavy on parsers and oracle tests.

Overview
FE-1253 reframes prospect research as a deterministic full-stack execution regression case, not a second end-to-end 2×2 campaign. Docs and D139-L record that the exploratory mission stays for specification work while study profiles, requirement ledgers, scored provider gates, and campaign claims are retired for this case.

A new frozen prospect-research-workspace-v1 packet fixes React/Node/TypeScript/SQLite, npm test / build / start, env-driven fixtures, forbidden runtime network, and accessible UI/API contracts. The compiled oracle (prospect-research-workspace-oracles-v1) runs npm test/build, then seven independent browser/HTTP/SQLite/export journeys against a controller known-good app and rival behavior modules (PR1–PR9).

Execution plumbing grows to four compiled oracles: strict public-contract parsing, closed oracle-pack hashing, operator CLI case resolution, Brunch workspace seeding via an opaque full-stack seed (not Petri section parsing), and dispatch/retain wiring in execution-comparison-operator. Tests lock registry membership, claim coverage, and rival sensitivity.

^{Reviewed by Cursor Bugbot for commit c0c3e47. Bugbot is set up for automated code reviews on this repo. Configure here.}