You make an interesting argument - and I'm not saying you're wrong.
It is not only an argument, this is more pending on the circumstance that this both
IDS systems are doing not the same thing!!!
One is watching and sniffing in the network or the network traffic it self and the other
one is watching on the host OS of an Server, PC or other devices watching their registry,
file system or other elementary or urgent points in that OS.
However, your definition of a "router" vs a "server" seems at odds.
What here should be better matching is perhaps something such as TripWire or something
else similar to that but not a Host IDS (HIDS). One thing is OS related and the other one is
network related or pointed. The Server has an OS that is perhaps hardened the firewall or
router OS (firmware) must be hardened.
For example: what is the difference between a pfSense device running an OpenSSH endpoint vs a server running the same thing?
pfSense is a firewall distribution (but here working likes a firmware of an network device) and let
us say CentOS & SoftEtherVPN are an OS & Software. Their fore what you are asking for should be
matching more well this software or perhaps able to realize combined installed on an appliance;
fail2ban
DenyHost
TripWire
How do you make the judgement in this case as to which device warrants an OSSEC agent?
Its not me, you should perhaps read the statements and jobs that the software coder where telling
their clients and perhaps too you could read about the differences NIDS and HIDS.
OSSec getting started