[pei711]

📝 Pull Request Template

1. Related Issue

Closes #642

2. Type of Change (select one)

Type of Change: Bug Fix

3. Description

In python/valuecell/server/services/conversation_service.py, get_conversation_list accepts limit and offset parameters forwarded from the GET /conversations endpoint. While total = len(conversations) was calculated, the conversation list was never sliced by [offset : offset + limit], causing the endpoint to return all conversations unconditionally regardless of requested pagination parameters.

This PR applies pagination slicing to conversations using [offset : offset + limit] (with non-negative guards) after strategy conversations are filtered, ensuring that:

1. Only the requested window of conversations is transformed and returned.
2. total still correctly reflects the total count of non-strategy conversations.
3. Requests with offsets beyond the total count return an empty list without error.

4. Testing

• I have tested this locally.
• I have updated or added relevant tests.

Added unit tests in python/valuecell/server/services/tests/test_server_conversation_service.py:

• test_get_conversation_list_applies_limit_and_offset: verifies correct window slicing and element order.
• test_get_conversation_list_offset_beyond_total: verifies graceful handling of offsets exceeding total count.
• test_get_conversation_list_excludes_strategy_agents_before_pagination: verifies filtering order relative to total count and pagination.

Ran full test suite (uv run pytest: 404 passed), code formatting (ruff format), and linting (ruff check: all checks passed).

5. Checklist

• I have read the Code of Conduct
• I have followed the Contributing Guidelines
• My changes follow the project's coding style