I can guess a reason: persistence of your IRC server connection(s), across device sessions, and maybe switchable between devices. Without using an IRC bouncer.

So this this would a turnkey way to run this somewhere centralized and persistent, and then you connect to it however you connect to that Docker container (e.g., SSH, remote desktop of some kind).

Of course, a non-Docker way to achieve simple persistence would be to just run a character-terminal IRC client in an SSH-able shell account (or VPS or AWS EC2), inside a `screen` or `tmux` session that can be detached and reattached when SSH-ing in from whatever devices.

(Persistence of your IRC server connections means things like you can see what you missed in scrollback, you aren't being noisy in your channels with join and part messages, you preserve your channel operator status and other channel modes without relying on bots, and you aren't leaking so much info about your movements in real time to random crazy people who hang out in Internet chat rooms.)

(Also, early on, if your leet channels attracted trolls, remaining connected meant that whatever automated countermeasures your client had could help defend the channel. Also, the more people who had channel operator status, the harder it would be for an attacker who, say, "netsplit" to hack ops, to de-op them all, before a remaining op's scripts detected the mass-deop attack, and took out the attacker. Also, your persistence bouncer or shell account obscured your real IP address, so if an attacker targeted your client's IP addr but not your home addr, such as with a protocol or flood attack, you could more likely get back on quickly. Trolls were often annoying, but it was also cyberpunk satisfying when your channel made short work of them.)

Back in the day I would keep an old PC running Linux in the closet just for my IRC and shell needs. Having a vanity domain name was a must if you were lucky enough to have a static IP. I remember undernet adding support to hide your IP once you created an account.