Unfortunately, top is a foreground process. And while such a foreground process is working, you can't really know whether the user is active or away. That is why any connection/shell timeout mechanics (ssh timeout, shell timeout, etc.) cannot work in this case.
The only way I can think of accomplishing what you (probably) want is to timeout the process/command itself. It works for a single and specific command, you'd need to execute this solution for any command you want to be timed out.
To timeout a command in a shell:
timeout 60 command
Where 60 is the time (in seconds) that the shell waits before forcefully stopping the process.
You could use the timeout command to avoid forgetting foreground processes like top open which leads to endlessly open SSH connections:
timeout 3600 top
Of course it is a nuisance to always write out the whole command with timeout every single time and you can easily forget to do it, so you can just alias the top command, or in another case - if you want it global, you can move the top binary somewhere else and replace /usr/bin/top with a script that starts the top command with a timeout. Note that if you want to use it in a script like that, you'd need to use the --foreground switch of timeout, otherwise it'll start the command in a subshell. Contents of /usr/bin/top for example:
/usr/bin/timeout --foreground 3600 /path/to/real/top
Another note: If you decide to use the "global solution", you should give it a little bit more thought. What I've given is simply a suggestion and a direction. The solution will stop working after a package update to top or it could even potentially break the update.
