I agree that this player FAICS was cheating. After all, is it not common sense to move into a spot where you don't run the risk of being attacked if you go AFK?
Additionally, there is the "/away <text>" command. If that was activated, then the GM would have known, because they send the suspects a /tell. And "/away <text>" replies to a /tell with "(auto-reply) <text>". So it was obviously not activated.
Yes, there always is the slim chance of going AFK due to urgent matters that don't even allow moving away. In these cases a client shutdown isn't always feasible, but that should be far rarer than usual, and the suspect didn't claim anything like this.
Thus it is conscious AFK training, and I don't see why that would not be treated just like external botting. After all, it is exploiting a bug (namely, lack of implementation).
The issue of some visual AFK indication, along with some "stasis", has been discussed before. The essential problem was the then possible abuse of it to get around dying, but that could be countered by activating the stasis for any MOB that would engage in combat after AFK has been turned on. This way, any fight that already started would continue as usual, but no new would start.
Unrealistic - yes. But it's an OOC solution to an OOC problem, so it doesn't qualify for realism, anyway. In fact, if we were to make sitting that marker, then there would be "RPers" who claim that sitting is a good way to hide...
Edit: I think it would be a good thing that whatever AFK method gets implemented, it should auto-activate after some time of inactivity (except a boot for reasons already stated). That won't prevent bots, but it would help legitimate players who unexpectedly go AFK. Maybe this feature could even be manually turned off, in the case of the RP-support alt doing the waiting routine. /Edit