Refactor: Move evaluation suite to dedicated evaluations module

[andrewginns]

This pull request reorganises the Mermaid evaluation tools by moving them from multi_mcp/eval_multi_mcp to a new dedicated module at evaluations/mermaid_evals. All corresponding references in documentation and build scripts have been updated to reflect this change, improving the project's structure and separation of concerns.

Problem:

The evaluation tools for Mermaid were previously located within multi_mcp/eval_multi_mcp, which mixed evaluation code with other unrelated components. This made it harder to maintain, understand, and extend the evaluation suite.

Solution:

The evaluation tools have been moved into a new directory, evaluations/mermaid_evals, to provide better separation between evaluation logic and other project components. All documentation and build scripts have been updated to ensure compatibility with the new structure.

Unlocks:

• Clearer project organisation, making it easier for contributors to locate and work with evaluation tools.
• Improved maintainability and extensibility for future evaluation features or frameworks.
• Reduced coupling between evaluation logic and other modules.

Detailed breakdown of changes:

• Moved Mermaid evaluation suite from multi_mcp/eval_multi_mcp to evaluations/mermaid_evals.
• Updated all references to the evaluation tools in documentation files.
• Modified build scripts to use the new evaluation tool paths.
• Refactored any import statements or configuration pointing to the old location.
• Cleaned up residual files or outdated references from the previous directory.
• 13 files changed in total, with 317 additions and 141 deletions.