Eliminate Snapshot Duplication in Session JSON

[CyberSecDef]

Problem

The session JSON file stores every piece of novel data twice:

1. Top-level keys — title, premise, genre, chapters, etc. (~199 KB)
2. progress_data.snapshot — An identical copy (~199 KB)

Two independent persistence paths evolved:

• save_session_state() writes top-level keys
• persist_completed_chapters() writes full progress_manager.get(token) dict (including snapshot)

Duplication doubles file size (~400 KB before chapters, ~1-2 MB for 25-chapter novel).

Files Affected

• novelforge/routes/generation/chapters.py lines 67-102
• novelforge/routes/generation/revision.py lines 78-92
• novelforge/routes/export.py lines 113, 138, 555-603
• novelforge/session/persistence.py lines 226-280, 329-351, 354-407, 471-512
• novelforge/progress.py

Why It Matters

• File size is 2x what it needs to be
• Two copies can drift if a bug updates one but not the other
• ~200 KB redundant data through read-modify-write cycle on every chapter completion

Recommended Approach

1. Stop persisting the snapshot — strip snapshot from progress data before writing
2. Rebuild snapshot on load from top-level keys
3. Eliminate completed_chapters / chapters_done duplication
4. Ensure every stage change persists
5. Backward compatibility for old session files