Fix D++ from stalling for a second before sending REST requests using poll.

[voltflake]

Quite a while ago i noticed an issue that makes windows D++ builds delay sending messages, responses, and other REST requests with 1000ms delay. This PR fixes that.

Basically i added a dummy socket into poll implementation so that when poll() is running and waiting for timeout it can be interrupted by event of that dummy socket and refresh its file descriptors on demand.

To properly test this, increase poll timeout to 5000 ms and try simple ping-pong example bot. Before this change bot would hang for 5s and only then reply. Or slightly faster if heartbeat event occurs, which triggers poll activation. Issue persisted on all platforms when using default poll socket engine, both Linux and Windows.

Feel free to pinpoint in flaws that i may missed or request changes.

• I have ensured that all methods and functions are fully documented using doxygen style comments.
  (they are as documented as they were before me)
• My code follows the coding style guide.
• I tested that my change works before raising the PR.
• I have ensured that I did not break any existing API calls.
• I have not built my pull request using AI, a static analysis tool or similar without any human oversight.

[netlify]

✅ Deploy Preview for dpp-dev ready!

Name	Link
🔨 Latest commit	ae95e14
🔍 Latest deploy log	https://app.netlify.com/projects/dpp-dev/deploys/68cd618239564400072e881b
😎 Deploy Preview	https://deploy-preview-1485--dpp-dev.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify project configuration.

[CLAassistant]

All committers have signed the CLA.

[braindigitalis]

what exactly is this doing? it looks like it continually inserts writeable udp fd's into the poll set?

[voltflake]

It inserts that wakeup udp socket into poll fds only once in constructor. Purpose of that socket is to allow poll() to exit before reaching timeout if needed. If other thread updates or registers new socket while poll() is already running and waiting for new socket events, it can't exit that state until timeout reaches or one of previously registered sockets receives event.

In other socket engine implementations, like epoll(), early exits are possible and are used by lib. This additional wakeup socket basically adds this interrupt functionality to default poll(), allowing it to exit early on demand.

refresh_pool() writes a byte to wakeup socket and drain_wakeup() discards that byte basically.

[Jaskowicz1]

I'm personally happy with this PR, but since I didn't work on the original implementation and I'm still a bit clueless on how the socket engine works, I'd rather @braindigitalis approve of this

[Jaskowicz1]

I've approved runners and updated the name, as this is reproduceable on Linux.