It seems like there has been a change internally to how the mime type of files is determined. I had a blob that returned a mime type of application/pdf and a file that was already an application/pdf, but got an error with Drive.Files.update, that it couldn't convert a "text/html" file to an application/pdf. But, it didn't need to convert it. They were both already application/pdf types according to the code. I think the problem was that the utilities service was not evaluating HTML tags, and leaving them in the content, and so the Drive.Files.update thought that the source blob was html even though it had a mime type of application/pdf.
The old code that worked probably had to many steps in it anyway.
I thought it might have been the V8, but changing back to Rhino didn't fix anything.
Basically I changed from:
//This worked previously but doesn't seem to any more
blob = HtmlService.createHtmlOutput(html).getBlob;//To test if the HTML is bad
pdfBlob = blob.getAs('application/pdf')
to:
pdfBlob = Utilities.newBlob(html, "application/pdf");