Replace ping sweep with direct parallel port scan across full subnet

Ping sweep + wait was the bottleneck: it blocked until all 254 pings
timed out before starting the port check. Now all 254 nc checks fire
simultaneously and we stop polling as soon as the first hit is written,
completing in ~1s instead of ~1 minute.

setup_onyx.sh

@@ -77,40 +77,25 @@ detect_ollama_host() {
     local LOCAL_IP
     LOCAL_IP=$(ipconfig getifaddr en0 2>/dev/null || ipconfig getifaddr en1 2>/dev/null || echo "")
     if [ -z "$LOCAL_IP" ]; then
-        echo "" >&2
         print_warning "Could not determine local IP address." >&2
     else
         local SUBNET
         SUBNET=$(echo "$LOCAL_IP" | cut -d. -f1-3)
 
-        # Ping-sweep the subnet in parallel to populate the ARP cache,
-        # then read ARP for live hosts and check port 11434 on all of them.
         echo "" >&2
-        print_warning "Ping-sweeping ${SUBNET}.0/24 to find live devices..." >&2
+        print_warning "Scanning ${SUBNET}.0/24 for Ollama (port 11434)..." >&2
-        for i in $(seq 1 254); do
-            ping -c1 -W1 "${SUBNET}.${i}" &>/dev/null &
-        done
-        wait
-
-        local ARP_IPS
-        ARP_IPS=$(arp -a 2>/dev/null | grep -v incomplete | grep -oE '\(([0-9]{1,3}\.){3}[0-9]{1,3}\)' | tr -d '()')
-
-        if [ -z "$ARP_IPS" ]; then
-            print_warning "No live devices found on ${SUBNET}.0/24." >&2
-        else
-            local COUNT
-            COUNT=$(echo "$ARP_IPS" | wc -l | tr -d ' ')
-            print_warning "Checking $COUNT live devices for Ollama (port 11434)..." >&2
 
         local TMPFILE
         TMPFILE=$(mktemp)
 
-            while IFS= read -r ip; do
+        # Launch all 254 checks simultaneously
+        for i in $(seq 1 254); do
+            local ip="${SUBNET}.${i}"
             echo -e "  ${BLUE}→${NC} Checking $ip..." >&2
             ( nc -z -w 1 "$ip" 11434 2>/dev/null && echo "$ip" >> "$TMPFILE" ) &
-            done <<< "$ARP_IPS"
+        done
 
-            # Stop as soon as the first result is written
+        # Poll every 200ms and stop as soon as the first result appears
         while [ "$(jobs -rp | wc -l)" -gt 0 ]; do
             if [ -s "$TMPFILE" ]; then
                 kill $(jobs -rp) 2>/dev/null || true
@@ -141,8 +126,7 @@ detect_ollama_host() {
             return
         fi
 
-            print_warning "Ollama not found on any live device." >&2
+        print_warning "Ollama not found on any device in ${SUBNET}.0/24." >&2
-        fi
     fi
 
     # Nothing found — ask the user directly