ci: revert #16249

[netrunnereve]

Since the AMD CI machines are running normally again we should be able to reenable them for all PRs.

[CISC]

@ggerganov I think the runners are stalled again. :(

[CISC]

..either that or just extremely slow:
https://github.com/ggml-org/llama.cpp/actions/runs/19427929770/job/55611134861

[ggerganov]

They accept SSH connections but any command that I run just stalls. Restarted them - not sure what is the problem.

[CISC]

@ggerganov Jobs are still not being picked up.

[ggerganov]

Forgot to start the GHA runner. Should be good now.

[netrunnereve]

Aaand it's dead again 😢

It looks like it stopped here.

[ggerganov]

Yes one of the machines was stuck again. Restarted

[CISC]

Is there anything that can be done to try and improve stability, or at least automatically restart when stuck? It got stuck again over the weekend...

[ggerganov]

Not sure - when these machines get stuck even just running simple ls command has no response. Strangely, it only happens with the NV4ads machines.