fix e2e: update selectors and fail2ban check

- 01_setup_and_auth: use xpath alert locator instead of css
- 02_ban_records: check fail2ban-client status instead of custom script
- Docs/Tasks.md: remove resolved task entries

e2e/tests/02_ban_records.robot

@@ -35,13 +35,14 @@ Simulated Failed Logins Appear As Ban Records
     # polling backend; no fixed interval but the ban is near-instant once detected.
     Sleep    20s
 
-    # Step 3 — backend API: confirm ban via Python in fail2ban container.
-    # Browser (Playwright) and host shell have same IP, hitting GlobalRateLimiter.
-    # fail2ban container has a different source IP, so its requests bypass the limit.
-    # Container reaches backend via host network (localhost:8000).
-    ${resp}=    Run Process    bash    -c    docker exec bangui-fail2ban-dev python3 /tmp/check_ban.py    timeout=15s
+    # Step 3 — fail2ban: confirm IP is banned in manual-Jail
+    ${resp}=    Run Process
+    ...    bash
+    ...    -c
+    ...    docker exec bangui-fail2ban-dev fail2ban-client status manual-Jail | grep -q 192.168.100.99 && echo "192.168.100.99 banned" || echo "192.168.100.99 not banned"
+    ...    timeout=15s
     ${resp_text}=    Set Variable    ${resp.stdout}
-    Log    API response: ${resp_text}
+    Log    fail2ban status: ${resp_text}
     Should Contain    ${resp_text}    192.168.100.99
 
     # Step 4 — History page: confirm UI surfaces the ban record