I agree that a good researcher, biographer, historian, will seek to go to the most primary version that can. However for example, I have access to hundreds of newspapers, the actual images of the actual columns from the time they appeared. Most people do not.
Sure they do. You just have to go down to the local library. For the most part, we're talking about academics/students so they can just go to the Uni library which will have plenty of old newspapers.
Now let's say I state "Henry Fonda still maintained relationship with his ex-wife Margaret Sullavan as they were seen eating lunch together months after the divorce" and I cite my source as the "Fresno Examiner", 4 Apr 1934.
Now someone could come along to my page, think that's interesting and cut and paste it directly into Wikipedia, obviously citing the newspaper but forgetting the courtesy of citing my work as the secondary citation. They did not actually read the newspaper, they are leaching off my work to present some interesting trivia to the world without even an acknowledgement.
That's backwards from what we're talking about. We're talking about Wikipedia being the intermediate source for an academic paper, not an academic paper being the intermediate source for Wikipedia. Regardless, it's a good example of why you shouldn't trust Wikipedia articles without verifying the source.
I try not to do that with my own sources, where I can't actually get a copy of the underlying source, and I wish others would make an effort to learn secondary citation. Aside from that it's sometimes rather important to know that a bit of data has been selected and filtered through an intermediary, sometimes that knowledge alone colors the reading.
You should always cite the source you used, that's just common sense. I can't see how anyone could argue that it's appropriate to cite something you haven't looked at. Whether you should include that source's source as well probably depends on the circumstances.