Original PDF object is being altered beyond adding an OCR layer.

I have not run the code, i just looked at the sample input and output files provided in the readme on the front page of this git.

Why does the input file go from 82k to 643k? Adding an OCR layer should not cause the file size to increase by almost 800%!

Taking a closer look the file itself is being altered, which i feel is unacceptable. All that should happen when creating a searchable pdf is adding a transparent text layer to the original pdf.

```
root@debian-test:~# pdfimages -list SampleInput.pdf
page num type width height color comp bpc enc interp object ID x-ppi y-ppi size ratio
--------------------------------------------------------------------------------------------
1 0 image 1267 793 rgb 3 8 image no 6 0 72 72 80.3K 2.7%
1 1 smask 1267 793 gray 1 8 image no 6 0 72 72 996B 0.1%
root@debian-test:~# pdfimages -list SampleOutput.pdf
page num type width height color comp bpc enc interp object ID x-ppi y-ppi size ratio
--------------------------------------------------------------------------------------------
1 0 image 5279 3304 rgb 3 8 jpeg no 6 0 72 72 642K 1.3%
```

Start with the sample input and output files linked from the README, then reproduce the conversion and compare both files with pdfimages -list. Trace where the output image dimensions, encoding, and size change; done means the searchable PDF preserves the original PDF content and adds only the transparent OCR text layer.

Written by the indexing model from the issue text.