In many of the PDFs my users are working with, copying text out of the PDF loses all whitespace, with the words all concatenated together. This makes the PDF Embed API unusable, and they're having to fall back to the browser's native PDF viewer.
In the paper available here (open access, PDF also attached), copying the caption for Table 1 in the Chrome PDF viewer gives "The acylcarnitine profiles in dried blood spots detected by tandem mass spectrometry", but the PDF Embed API gives "Theacylcarnitineprofilesindriedbloodspotsdetectedbytandemmassspectrometry" instead, which is not useful for my purposes. This is particularly common with table and figure captions.
Sorry for the inconvenience. We have noted this down and will let you know when it gets resolved. Stay tuned!