fix: correct MCPServer call_tool result type

[fengjikui]

Summary

Fixes #2695.

MCPServer.call_tool() currently advertises Sequence[ContentBlock] | dict[str, Any], but the convert_result=True path never returns a raw dict. It can return only:

• a direct CallToolResult,
• a Sequence[ContentBlock] for unstructured tools, or
• a (content, structured_content) tuple for structured tools.

This PR removes the unreachable raw-dict branch in _handle_call_tool(), drops the now-unused json import, and introduces a ToolResult type alias so the public annotation and handler implementation share the same result shape.

Tests

• uv run --frozen pytest tests/server/mcpserver/test_server.py -q
• uv run --frozen pytest tests/server/mcpserver/test_server.py::TestServerTools::test_call_tool_returns_declared_result_shapes -q
• uv run --frozen coverage erase && uv run --frozen coverage run -m pytest tests/server/mcpserver/test_server.py && uv run --frozen coverage combine && uv run --frozen coverage report --include='src/mcp/server/mcpserver/server.py' --fail-under=0 && UV_FROZEN=1 uv run --frozen strict-no-cover
• uv run --frozen ruff format --check .
• uv run --frozen ruff check .
• uv run --frozen pyright
• uvx --from pre-commit pre-commit run markdownlint --files docs/migration.md
• uvx --from pre-commit pre-commit run end-of-file-fixer --files docs/migration.md src/mcp/server/mcpserver/server.py tests/server/mcpserver/test_server.py

AI assistance disclosure

I used an AI coding assistant to help trace the result shapes and draft the patch. I reviewed the final diff and ran the validation commands above locally.