fix: increase delay for IP detection to reflect connection state changes

The WARP daemon requires ~5 seconds to fully process connect/disconnect commands and update the routing. Previously, 1500ms was insufficient, causing IP addresses to show as unchanged. Increased delay to 5000ms (5 seconds) to ensure IP queries reflect the actual connection state.

Testing shows:
- warp down: Before WARP IP → After ISP IP ✓
- warp up: Before ISP IP → After WARP IP ✓

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>

Author: ashu17706

src/commands/connect.rs

@@ -1,6 +1,8 @@
 use anyhow::Result;
 use serde_json::json;
 use colored::Colorize;
+use std::thread;
+use std::time::Duration;
 use crate::format;
 use crate::warp_cli;
 use crate::ip;
@@ -12,6 +14,9 @@ pub fn run(connect: bool, json: bool, quiet: bool) -> Result<()> {
 
         warp_cli::connect()?;
 
+        // Wait for connection to take effect
+        thread::sleep(Duration::from_millis(5000));
+
         // Get IP after connecting
         let after_ip = ip::get_public_ip().ok();
 
@@ -42,6 +47,9 @@ pub fn run(connect: bool, json: bool, quiet: bool) -> Result<()> {
 
         warp_cli::disconnect()?;
 
+        // Wait for disconnection to take effect
+        thread::sleep(Duration::from_millis(5000));
+
         // Get IP after disconnecting
         let after_ip = ip::get_public_ip().ok();