I don't think the problem necessarily is that Tesseract is slow, but that the whole process of rendering a PDF to a series of PNGs on which you can then run OCR is slow (which is what it does in the background).
The process of converting all pages to raster images and then OCR-ing each one takes hours for PDFs hundreds of pages long. This workflow is not suitable for instant search. For non OCR-ed PDFs it's worth to pregenerate the text.
That's why rga comes with a cache. I've occasionally used the Tesseract adapter with good success (results-wise), and after the inital rendering and indexing, it's fast enough to use.
