Feature: Split ExplorePage into tab and section modules (EPIC-3.5)

[masvelio]

Summary

Decompose ExplorePage into a stable tab shell plus clearer section modules for feature-owned discovery content.

Problem / Motivation

ExplorePage is smaller than the thread/discussions pages, but it now carries prediction-market discovery UI in addition to quests and other discovery sections. Keeping it as a single container slows later feature migration and keeps ownership unclear between the page shell and section-level features.

Proposed Solution

• Split ExplorePage into a tab shell and explicit feature-owned sections.
• Treat prediction-market discovery surfaces as first-class scope, not incidental UI.
• Isolate section-specific state, filters, and list wiring from the page shell where possible.
• Preserve current tab navigation, search/filter behavior, and route behavior.

Acceptance Criteria

• ExplorePage is decomposed into a tab shell plus section-level modules.
• Prediction-market discovery components are no longer embedded as ad-hoc page internals.
• Tab switching, search, and filter behavior remain unchanged.
• The resulting structure is ready for EPIC-4.15 feature migration.
• Build/typecheck and relevant component/E2E checks pass.

Alternatives Considered

• Defer all Explore cleanup to EPIC-4.15. This was rejected because the page now contains cross-feature discovery UI that should be normalized before migration.
• Keep prediction-market discovery inside the page while splitting only the tab shell. This was rejected because it leaves the newest high-risk surface unowned.

Impact

This reduces migration risk for discovery surfaces, especially prediction markets, and creates a cleaner Explore shell for later feature moves.

Links

• Plan doc: common_knowledge/Commonwealth-Frontend-Refactor-Plan.md
• Milestone: https://github.com/hicommonwealth/commonwealth/milestone/136?closed=1

Additional Context

The latest plan audit explicitly pulls prediction-market discovery into EPIC-3/4 scope. Keep this issue focused on page normalization and ownership boundaries rather than a full feature move.