fix bag: Add OCR file purpose and update OCR example

[suekou]

Due to OCR not functioning in the SDK, I corrected the problematic section. I tested the modified part and confirmed that it works properly.

Modifications:

• Added OCR to the purpose.
• Fixed a typo in the example file. Consequently, the documentation was also updated PR.

[suekou]

Additional context:
According to this documentation, the 'purpose' is expected to be "ocr". However, in the actual implementation, it was not possible to assign "ocr" to the purpose field, so I added OCR: "ocr" to the FilePurpose.

Also, as you can see here, the expected field is includeImageBase64 rather than include_image_base64. Therefore, I have updated the relevant example accordingly.

[vicentematus]

I was confused too when I implemented the Mistral AI SDK, because using purpose:'ocr' the language server was mad at me:

Example of the code snippet to reproduce it:

const uploaded_file = await Bun.file("example.pdf").arrayBuffer();
const uploaded_pdf = await mistral.files.upload({
  file: {
    fileName: PDF_NAME,
    content: uploaded_file,
  },
  purpose: "ocr",
});