Fix GPU device options missing from module/device selectors

[thomaswantstobeaskeleton]

Motivation

• Module/device pickers were only showing Default and cpu because runtime device discovery was fragile and didn’t expose GPU device options in AVAILABLE_DEVICES.
• The goal is to harden device detection so GPU runtimes (including indexed CUDA devices) appear in module selectors and the default device picker.

Description

• Centralized runtime device option population in modules/base.py by introducing _append_unique_device and building AVAILABLE_DEVICES from a single source of truth.
• Improved CUDA/ROCm detection by checking both torch.cuda.is_available() and torch.cuda.device_count() and adding cuda plus indexed entries (cuda:0, cuda:1, ...) when present.
• Preserved and deduplicated detection for XPU, MPS, and DirectML (privateuseone) while keeping the CPU fallback intact.
• DEVICE_SELECTOR() continues to read from AVAILABLE_DEVICES, so UI selectors and module defaults automatically pick up discovered devices.

Testing

• Ran python -m compileall -f -q modules/base.py and it completed successfully.
• Performed a runtime smoke check with python - <<'PY' ... import modules.base as b ... PY that printed DEFAULT_DEVICE, AVAILABLE_DEVICES, and DEVICE_SELECTOR(); the script ran successfully and showed CPU-only results in this environment (expected because no CUDA devices are exposed here).
• No automated test suites were modified or run beyond the compile and runtime smoke check described above.

---

Codex Task

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 435456aed3

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Require CUDA availability before choosing it as default

The new condition torch.cuda.is_available() or cuda_count > 0 can set DEFAULT_DEVICE to cuda even when PyTorch reports CUDA is not usable in the current process. This happens in partially broken CUDA installs (e.g., driver/runtime init failures) where device enumeration can still return a positive count; the app will then default to CUDA and later fail on model load or cache ops instead of safely staying on CPU. The previous behavior only selected CUDA when is_available() was true, so this introduces a regression in startup/device-selection reliability.

Useful? React with 👍 / 👎.