Docker and all the other fancy names are just tools. The examples are just examples.
Resource-wise, Docker changes nothing much beyond usual Linx/Unix multiprocessing. It does not have the overhead of VMs, and little overhead over just starting the process directly ( https://stackoverflow.com/questions/21889053/what-is-the-runtime-performance-cost-of-a-docker-container ).
So it is really up to you to decide how to use it. If you have closely related applications under your own command, maybe in your intranet, why not use a shared DB. If they are customer DBs, or you might need to run different versions of the software, or different incompatible configurations, then separate services might be fine.
For RDBMSses especially, they are usually very good at managing resources (RAM, HDD) and so for this case specifically, running only one may just be the best thing. It's really more about what your needs are than general guidelines.