coretasks: join channels on MOTD

[Exirel]

Description

Tin, should fix #2289 for good but I didn't test that section in particular.

This was way too easy. The tests didn't even fail, everything worked on the first try. I don't really complain, I'm just... yeah, OK, all good then.

I've tested it with the Libera servers (they use solanum-1.0-dev), and with a friend's server using UnrealIRCd-5.2.4. Both are fairly modern IRC servers, so it is possible that this breaks with older server that don't follow the MOTD specification.

Checklist

• I have read CONTRIBUTING.md
• I can and do license this contribution under the EFLv2
• No issues are reported by make qa (runs make quality and make test)
• I have tested the functionality of the things this change touches

[half-duplex]

This fixes the specific issues I was observing in #2289. I haven't poked into the state graph - is this still a race, just one we more reliably win? Or do we know that we'll never get an ENDOFMOTD/NOMOTD until after the UHNAMES has been received?

[dgw]

Horse docs says that:

Upon successful completion of the registration process, the server MUST send, in this order, the RPL_WELCOME (001), RPL_YOURHOST (002), RPL_CREATED (003), RPL_MYINFO (004), and at least one RPL_ISUPPORT (005) numeric to the client.

The UHNAMES flag, if advertised, will appear in a 005 line, at which point Sopel will immediately send the PROTOCTL to enable it.

Some other stuff "SHOULD" or "MAY" happen before the server again "MUST" respond as if the client sent it the MOTD command, which will trigger the start of channel joins as of this patch. By that time, since coretasks stuff tends to use plugin.thread(False) to guard against processing protocol events in parallel, PROTOCTL UHNAMES should have taken effect.

Is it technically still a race? Yeah, I guess. But by the time Sopel processes everything the server has sent and starts asking to join channels, a race should be practically impossible because disabling threaded handling forces everything the server sends after registration to be processed in sequence.

[Exirel]

So, is that an approval?

[dgw]

Ask @half-duplex, as the original reporter of #2289.

[half-duplex]

Even if there's still potentially a race, between the try/catch, this, and that bad results don't actually matter, I'm fine calling it resolved.