feat: Mistral ocr instrumentation

[CakeCrusher]

Video showing it (skip the intro):
https://www.youtube.com/watch?v=Bx1oJBlRfTY

Added Mistral OCR instrumentation for the .process method

• Followed existing mixin patterns to destructure the instrumentation into creating a _ocr_wrapper.py leveraging new extraction functions from _request_attributes_extractor.py and _response_attributes_extractor.py.
• Added tests verifying the correct span item are extracted from calling .process
• Added a tutorial showcasing it (with a lot of print statements)
• Ensured that the new instrumentation is backwards compatible

Note: I am aware its a lot of code, much of it is generated, but its following the current mixin pattern of instrumentation as used elsewhere in the repo. Also I think this feature is pretty horizontal so the added complexity I would not expect to contaminate the rest of the project.