I believe if there was any idle time then that should be registered in the base hit/operation.
Let me add some documented information about the server time. In releases 12.2 and earlier, server time was calculated only for the base hit. Now it is calculated for all hits in the operation. Calculating server time for all hits in the operation enables us to provide a more accurate and refined value. The Network time value might be lower, because the redesigned server time refers to the fragments of operation that were previously attributed to network time. Network time is part of the operation time in which all the hits in simultaneous sessions were recognized either as request time or response time.
The table is correct. It is Oracle server that was processing the transaction nearly all the time.
The chart is also correctly reflecting what is going on from the wire perspective. What the chart is not telling is what those hits exactly mean: each of them is a request to the server asking "hey server, do you have the answer already?". Server waits a while to see whether background thread completed user transaction (this red bar is visible) and if not, it responds "no, wait a second more". So the client waits and asks the question again. Situation repeats itself, with server instructing the client to wait more with every subsequent poll. Note the regularity of gaps between those bars. Finally when the answer is ready, server replies with the processed content.
What's missing in the chart is a clear indication that in this case the white spaces are not the client idle time. We will fix it in DC RUM release 18.
Small supplement to Kris's note: please, take a look at attached pdf files showing example report before and after update. With Server wait time metric shown on stepchart, it will be much easier to understand Server time value calculated for whole operation.