fix(chat_app): honor backend switch and context auto mode

[leehack]

Summary

• Align chat runtime backend status with selected backend so CPU mode is reflected correctly and GPU indicators do not report false positives.
• Preserve configured GPU layers when switching backend preference, letting load-time backend logic enforce CPU offload behavior.
• Restore Context Size Auto mode by allowing/saving 0 in settings and using engine-derived context limits during session management.
• Add regression tests for backend switching behavior and context-size auto persistence.

Testing

• flutter test test/unit_test.dart test/settings_service_test.dart
• flutter analyze

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 78.26%. Comparing base (72848b2) to head (e163fb5).

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #48   +/-   ##
=======================================
  Coverage   78.26%   78.26%           
=======================================
  Files          60       60           
  Lines        4637     4637           
=======================================
  Hits         3629     3629           
  Misses       1008     1008           

Flag	Coverage Δ
unittests	78.26% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.