fix: make HTTP check authoritative in healthcheck for container compatibility

Process detection fails in containers due to missing privileged tools
(ss -p requires root, lsof often not installed). HTTP health endpoint
is the reliable indicator of service health, so port/process checks
are now informational only.

scripts/healthcheck.sh

@@ -56,7 +56,7 @@ check_port() {
         fi
     else
         health_log "WARN" "Neither ss nor netstat available for port checking"
-        return 1
+        return 0
     fi
 }
 
@@ -108,11 +108,10 @@ check_process() {
             local cpu_time=$(awk '/utime|stime/ {sum+=$2} END {printf "%.2f seconds", sum/100}' "/proc/${pid}/status")
             health_log "INFO" "  CPU time: ${cpu_time}"
         fi
-        return 0
     else
-        health_log "ERROR" "✗ ${service} process NOT found"
-        return 1
+        health_log "WARN" "⚠ ${service} process NOT found (may require privileges)"
     fi
+    return 0
 }
 
 # System resource check
@@ -160,17 +159,13 @@ main() {
 
     # Check chat service (port 4500)
     health_log "INFO" "=== Chat Service (port 4500) ==="
-    if ! check_port 4500 "chat service"; then
-        exit_code=1
-    fi
+    check_port 4500 "chat service"
 
     if ! check_http "http://localhost:4500/api/health" "chat service" 3; then
         exit_code=1
     fi
 
-    if ! check_process "chat service" 4500; then
-        exit_code=1
-    fi
+    check_process "chat service" 4500
 
     health_log "INFO" "========== HEALTH CHECK END ==========="