…nalyzer from a simple client-side tool into a robust, extensible, and scalable platform.

The changes are implemented in three main phases:

Phase 1: Foundational Improvements
- Implemented session persistence using localStorage to prevent data loss on page refresh.
- Expanded the list of metric presets to include common defaults for PyTorch, Keras, and Hugging Face.
- Significantly increased unit test coverage for the core data parsing logic (`ValueExtractor`).

Phase 2: Platformization with a Backend
- Integrated the Firebase SDK to provide backend services.
- Implemented user authentication with Google Sign-In.
- Added a Firestore database backend to save and load workspaces for authenticated users, while retaining localStorage for anonymous sessions.

Phase 3: Extensibility with a Plugin Architecture
- Refactored the data extraction logic into a modular plugin system.
- Each extraction method (Keyword, Regex) is now a self-contained plugin with its own logic and UI configuration.
- A central plugin registry allows for easy addition of new parsing capabilities in the future.