Refactor ChEBI fetcher for new API and batch support

Replaces XML-based models and parsing with Pydantic models for the new ChEBI REST API. Adds support for batch fetching and searching of ChEBI entries, updates error handling, and refactors data mapping to the PyEnzyme v2 data model. Removes legacy XML parsing and streamlines the code for modern API usage.

Author: JR-1991