feat: optimize context compression and session sync (#402)

* feat: optimize context compression and session sync

Context Compressor:
- Add ContentBlock type support for user messages with images/files
- Enhance buildConversationHistory to handle ContentBlock[] format
- Update serializeForSummary to convert ContentBlock[] to text
- Support mixed content (text, images, files) in history compression

Session Sync:
- Replace loop of addMessage with batch addMessages (transaction-protected)
- Use addMessages for atomic bulk insert with BEGIN/COMMIT
- Ensure data consistency before calcAndUpdateUsage reads DB

Chat Run Socket:
- Remove convertHistoryFormat before compression (compressor handles raw format)
- Compressor now processes original messages for better tool message handling
- Improve compression quality by preserving original message structure

Benefits:
- Better compression quality with proper tool message recognition
- Improved performance with batch DB inserts
- Strong data consistency with transaction protection
- Support for rich content (images/files) in conversation history

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* chore: bump version to 0.5.7

- Prepare for release with context compression and session sync optimizations

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

* chore: add 0.5.7 changelog entries

- Optimize context compression to support rich content (images, files)
- Improve session sync with batch inserts and transaction protection
- Fix usage.updated event reception for accurate token tracking

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>

package.json

@@ -1,6 +1,6 @@
 {
   "name": "hermes-web-ui",
-  "version": "0.5.6",
+  "version": "0.5.7",
   "description": "Self-hosted AI chat dashboard for Hermes Agent — multi-model (Claude, GPT, Gemini, DeepSeek) web UI with Telegram, Discord, Slack, WhatsApp integration",
   "repository": {
     "type": "git",