Propagate errors instead of swallowing them in portal and installer scripts

[devin-ai-integration]

Summary

Every entrypoint in this repo had at least one path where a failure produced no signal: the portal swallowed clipboard errors with catch{}, the PXE installer downgraded failed boot-file downloads to a WARN and exited 0, and the Proxmox installer hid --reapply failures behind >/dev/null 2>&1 || true. This makes those failures visible and fatal where the result would otherwise be silently broken.

brand.js

• catch{} on the copy button → logs and shows Kopieren fehlgeschlagen; also detects the common real cause (navigator.clipboard unavailable over plain HTTP).
• Manifest loading: 15s timeout via AbortSignal.timeout, Array.isArray(data.brands) validation, and the concrete reason (HTTP status / JSON error / timeout) is rendered into the error card instead of only console.error.
• Rendering no longer throws on a missing DOM node or a malformed manifest entry (a missing #brand-grid used to abort loadManifest and surface as "manifest not available"); missing elements are reported once at load, and render failures fall back to the error card.

scripts/build-assets.py

• Added BuildError plus a main() wrapper, so a broken brand fails with ERROR: asset build failed: [bnd.zone/corporate] Canonical SVG source missing: ... and exit 1, instead of a raw traceback from cairosvg or a KeyError: 'tagline'.
• Validates brand.json (valid JSON, required brand/tagline), SVG source existence, accent hex format, and DejaVu font availability.

proxmox/install-webowie-proxmox-branding.sh

• Assets are downloaded to a mktemp -d stage and only installed into $INSTALL once all six succeed; previously a mid-loop curl failure left $INSTALL half-overwritten, breaking the next --reapply.
• [[ -e ... ]] && install ... as the last statement of the --reapply branch made the script exit 1 whenever dd_logo.png was absent → converted to if/fi.
• The HTML injection is verified: no </head> used to mean sed silently did nothing and the install still reported success.
• The apt post-invoke hook now appends to /var/log/webowie-pve-branding.log (it still can't fail dpkg) instead of discarding all output, and --reapply/--install check that staged assets exist and are non-empty.
• Refuses to self-install $0 when piped from curl (it would install the shell, not the script), and reports a failed systemctl restart pveproxy.service.

scripts/create-netbootxyz-alpine-lxc.sh

• CTID="${CTID:-$(pvesh get /cluster/nextid 2>/dev/null || echo 180)}" silently used CTID 180 when pvesh failed; now it fails with instructions.
• Boot-file downloads are split into ESSENTIAL_FILES (UEFI + legacy BIOS entrypoints) and optional ones: missing essentials abort, missing optionals are summarized. Failed downloads no longer leave truncated files in the webroot.
• pct start is followed by a wait for status: running and for pct exec readiness, rather than sleep 5 and a cryptic downstream failure.
• Empty IP detection (host and container) is reported instead of writing HTTP: http:/// into /etc/motd; update-ca-certificates || true now warns.
• Both scripts got an ERR trap that names the failing line and command.

Verified: python scripts/build-assets.py regenerates all assets byte-identically to main and error paths return BuildError; bash -n + shellcheck clean on both shell scripts (including the container heredoc checked as sh); node --check brand.js.

Link to Devin session: https://app.devin.ai/sessions/2249710876544504a25ff092b69ab79b
Requested by: @puchadave

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring