fix(setup.sh): guard SIGPIPE in site-agent pod lookup (spurious [7h] Code 141) (#2676)

## Problem
In the `[7h]` site-agent verification, the pod lookup is:
```
_POD="$(kubectl get pods -n nico-rest -l ... -o name 2>/dev/null | head -1)"
```
The script runs under `set -euo pipefail` (line 77). `head -1` closes
the pipe after the first line, kubectl receives **SIGPIPE → exit 141**,
and `pipefail` propagates it, so the command substitution fails and the
phase aborts with `SETUP FAILED ... Code: 141` — **even though
site-agent deployed fine** (StatefulSet ready, Site CR
`HandshakeComplete`). Reproduced on a clean v0.10.3 install; same code
on `main`.

## Fix
Append `|| true` so the SIGPIPE-induced 141 is tolerated.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Signed-off-by: kirson-git <ekirson@nvidia.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Co-authored-by: Hasan Khan <hasank@nvidia.com>

Author: 3 people

helm-prereqs/setup.sh

@@ -796,7 +796,7 @@ _CONNECTED=false
 for _i in $(seq 1 24); do
     _POD="$(kubectl get pods -n nico-rest \
         -l "app.kubernetes.io/name=nico-rest-site-agent" \
-        -o name 2>/dev/null | head -1)"
+        -o name 2>/dev/null | head -1 || true)"
     if [ -n "${_POD}" ] && \
        kubectl logs -n nico-rest "${_POD}" --since=5m 2>/dev/null \
            | grep -q "NicoClient: successfully connected to server"; then