[Bug]: No errors and no output for large DPI files about ocrmypdf

[dan-ryan]

I believe it's an out-of-memory issue. But the app should warn about it if possible.

[jbarlow83]

Most pages of this document have a small, very high resolution image in the bottom left corner. OCRmyPDF has to render the whole page at the same resolution for OCR, so it uses a heuristic to decide how to deal with small features that amounts to weighted average of the area occupied by the image.

There was an error in the calculation that tries to find an appropriate weighted DPI for OCR, so it put too much weight on that feature, leading to rendering everything at high resolution and running out of memory.