You are mostly on target. At some point the documents are scanned and converted via OCR into a searchable text file (this is actually somewhat cumbersome and unnecessary, but change in the mechanics of the legal system is glacial, at best). Both electronic and paper documents are involved. OCR still has issues with fonts that are complex in some fashion (bold, serifs, cursive, even mixed number-letter formats).
The initial filing was likely scanned from paper into an image, then processed via OCR back into a searchable electronic format (PDF) for transmission to the court. The probably immaculate paper documents will follow along and possibly never be looked at in more than a cursory fashion until they are used for trial prep.
I used to do a lot of work with electronic documents and document automation and much of it was for legal documents (lawyers). They submitted documents electronically, but they were followed up by paper copies.
Why paper? I'd expect the docs would go from the word processing software printed directly to pdf. Then the pdf would be uploaded. Generally, government agencies and courts accept ONLY pdf documents.
Now, I don't know if this is relevant - While some here have said that only adobe has a pdf engine, that's not true. There are others, including open source. Some of these will handle complex or unusual formatting differently.
But I will admit, the errors look like OCR errors. Assuming it was a collaborative effort, I'm not sure how the OCR came into play. Perhaps members of the team were using different software - Word for some, an Apple program for others, etc. At the last minute they may have tried to pull them together, but the conversions were not going well, so they used OCR to bring text from one document to another - copy the text you want to use, then run that through an OCR machine. I do that from time to time - I'll use OCR on some electronic text when I am trying to bring it into one of my docs. No paper is involved.
Any idea why the GA filing has no case number?
The initial filing was likely scanned from paper into an image, then processed via OCR back into a searchable electronic format (PDF) for transmission to the court.
No, it was converted to PDF electronically with no scanner.
The metadata shows: "xmp:CreatorTool: Acrobat PDFMaker 20 for Word"
https://helpx.adobe.com/acrobat/using/creating-pdfs-pdfmaker-windows.html
The whole source file is converted to a PDF file very quickly by pressing a button.