Automatically configure Labdata API during Pyodide initialization to provide
seamless access to internal test data platform.

Key changes:
- Add Labdata wheel package loading in bootstrap process
- Configure LABDATA_TOKEN environment variable in Python runtime
- Add pandas and requests as core dependencies
- Implement automatic retry logic for API requests

Technical details:
- Load labdata_api-0.0.12-py3-none-any.whl after Pyodide initialization
- Set LABDATA_TOKEN in Python environment for internal authentication
- Pre-install pandas and requests for data handling
- Add error callbacks for package loading diagnostics

Example usage:
```python
from labdata_api import Labdata
labdata = Labdata()  # No token needed, configured automatically

Add proper wheel build configuration for labdata_api package and comprehensive
documentation for wheel management.

Changes:
- Add pyproject.toml with proper build configuration
- Update wheel file for frontend/Pyodide integration
- Add detailed wheel management documentation

Technical details:
- Configure hatchling as build backend
- Set proper package includes/excludes
- Add dependency specifications
- Document wheel building, testing, and SHA256 verification process

 use an asynchronous initialization pattern with `httpx`  instead of `requests` library for API calls, exhibited critical failures when executed within a Marimo notebook environment. Specifically :
*   API calls intended to fetch a list of programs resulted in incomplete data being received.
*   This manifested as a `requests.exceptions.JSONDecodeError: Unterminated string...` because the received data was truncated mid-response.
*   Debugging confirmed a mismatch between the `Content-Length` header reported by the server (e.g., `112131` bytes) and the actual number of bytes received by `requests` within Marimo (e.g., `7746` or `32383` bytes).
*   Crucially, the *exact same code* using `requests` executed successfully when run outside Marimo (e.g., in a standard Python script), receiving the full response.