Add post-reveal explanation framing condition

[Spbd1]

Motivation

• Introduce a randomized framing manipulation for the post-reveal open-text explanation so participants are assigned to either explain-to-self or explain-to-other and wording can be adjusted without changing the study’s neutrality.

Description

• Add explanationFrameCondition?: { condition: "explain-to-self" | "explain-to-other"; assignedAt: string } to the ResearchSession model and an ExplanationFrameConditionName type in types/research.ts and include it in exports.
• Persist and normalize the field via utils/session.ts and expose assignExplanationFrameCondition(session) which assigns the condition once (no re-randomization) when the post-reveal survey starts.
• Update PostRevealSurveyForm to call assignExplanationFrameCondition at survey start and to show condition-specific, neutral wording for the open-text question while preserving the existing 10–500 character validation.
• Include the frame in computed outputs and exports by adding explanationFrame to computed metrics, adding explanationFrameCondition to the Zod researchExportSchema and CSV flattening (explanation_frame_condition), bumping the research schema version to hidden-cost-game-research-schema-4, and updating the sample export.

Testing

• Ran npm run typecheck and it completed successfully.
• Ran npm run lint; linting passed but reports an existing react-hooks/exhaustive-deps warning in ParticipantBackgroundForm.tsx (unchanged by this PR).
• Ran npm run build and the production build completed successfully.
• Ran npm run validate:sample to validate the sample export against the updated schema and it succeeded.

---

Codex Task