Great! I´ll review this bit-by-bit. One thing I'm concerned about in general terms is whether we're expecting microdata for multiple items on a page. My understanding of the format is that when I, e.g., cite a book in an article I could use this to embed citation information for the book.
I guess three questions:

1. Am I understanding that correctly?
2. If so, is that actually done to any meaningful degree?
3. If both 1. and 2., how are we thinking of handing that?

Yes, that is possible (but I haven't seen citations encoded in microdata so far)

My test cases for developing this translator were the translators already using itemprop somewhere, i.e. https://github.com/zotero/translators/search?utf8=%E2%9C%93&q=itemprop&type=. All of them should work fine with the current code and the microdata should improve some data. There is one exception for https://www.kitapyurdu.com/kitap/makroekonomi/139156.html where indeed several publication items are described in that page (main entry and related items). I categorized this as an exception and was happy that it works fine for all the other cases.

The detection of EM is not changed and therefore the microdata does not influence the detection in EM. The cases were it could go worse than before are therefore cases were we already now detect a single item but the microdata there now correspond to multiple items.

What may be really tricky is the creation of the itemid. The main item has the url as the itemid and other items (e.g. person) have another itemid. We could try to create different itemids for multiple publications which would automatically lead to different items outputed in the doWeb. But can we decide this in a generic way?

(There are two debug statements which you can comment out, to see some more information about the statements and the types.)

This needs further testing. I see some strange behavior e.g. in Springer translator now, which depends on EM.

@zuphilip interesting what was the problem with Springer? I'm trying to understand what potential issues we can possibly encounter by adding support for more metadata types to EM translator.

@mrtcode Sorry, I can't remember. But one can try to check out the branch and test the detection/extraction with the Springer website again.