Zipstack/llm-whisperer-python-client ? reverse-engineered prompt

Build me a small Python client library for the LLMWhisperer API v2 that makes it easy to send documents and get back extracted text in a way that works well for LLMs.

I want it to feel simple to use, with one main client class, clear methods for uploading or processing a document, and good support for both native text and OCR style documents like PDFs and images. It should handle the common cases cleanly, including layout preserving output and the special document form elements like checkboxes and radio buttons.

Please include solid error handling, a simple way to pass in an API key, and a few usage examples in the README. Also add tests, including unit tests and integration tests that can run when a real API key is provided. If any API details are unclear, look up the current docs online and match the latest v2 behavior.