We should not track in-correct results. But I also don't have strong feelings about the above mentioned corner cases.
- Opened in a new tab would be good to have (for completeness)
- New window is not relevant on mobile, isn't it? 
- Offline page and downloaded file should probably not be tracked with the same histogram since it's a different context and just pollutes the results. If you really want to track it, it should be separated

Opening in a new window in relevant on mobile in android N in split-screen mode.

The issue isn't that we don't track opening in a new tab or that we track it incorrectly. The issue is that however we open an article (same tab, new tab, etc) we track the time from opening the article until the current tab is closed as a single metric.

This works for opening in the same tab, but if the user was to open an article in a new tab we would record the time from then until the New Tab Page (which the user may just leave open in the background) is closed and this would be part of the histogram recorded.

This may get worse if we assume certain users will go through the list of articles, open multiple in new tabs and then deal with them later.

However in practice I'm not sure how likely this case is, and if it could be solved by just ignoring the outliers in the histogram

Merging into the newer bug which documents the current state of affairs.