These are excellent questions!
From my point of view, speaking and reading are a little different. When speaking conversationally, native speakers often take shortcuts, but when someone is reading, such as in an audio book for your entertainment, they are going to take the time to enunciate words. Audio books are a form of entertainment, so it is important that the reader paints a picture with his words. In your first example of "deep and green", the reader would want the listener to picture the Salinas River dropping into it's banks which run both deep and green. The best way to get you to visualize that is to slow down and enunciate. If I was just talking to you normally, I might say, "deep-n-green". In your second example, I would, if I were speaking conversationally, probably say withevery making that a hard "th". I would not say about "devery". I might use that if the word preceding every were "and".
Here is a good link that might help explain this to you a little better: https://www.gonaturalenglish.com/connected-speech-fast-native-english-pronunciation/
