Add per-model response length overrides - #1144

Summary

• Add optional responseLength normalization for multi-endpoint model rows and preserve compatibility with existing endpoint records.
• Add an admin model-endpoint UI field for per-model response length and serialize it with endpoint model settings.
• Apply the selected model's response length to standard chat completions with max_completion_tokens for GPT-5/o-series aliases and max_tokens otherwise.

Validation

• python -m py_compile application\single_app\functions_settings.py application\single_app\model_endpoint_clients.py application\single_app\route_backend_chats.py functional_tests\test_model_endpoint_normalization_backend.py functional_tests\test_streaming_multi_endpoint_resolution.py functional_tests\test_model_endpoint_protocol_inference.py functional_tests\test_model_endpoints_api_key_manual_models.py ui_tests\test_model_endpoint_request_uses_endpoint_id.py
• node --check application\single_app\static\js\admin\admin_model_endpoints.js
• python functional_tests\test_model_endpoint_normalization_backend.py
• python functional_tests\test_streaming_multi_endpoint_resolution.py
• python functional_tests\test_model_endpoint_protocol_inference.py
• python functional_tests\test_model_endpoints_api_key_manual_models.py
• python functional_tests\test_model_endpoint_model_icon_picker.py
• python functional_tests\test_model_endpoints_save_button.py
• python -m pytest ui_tests\test_model_endpoint_request_uses_endpoint_id.py -q (skipped without UI env)
• git -c core.whitespace=blank-at-eol,blank-at-eof,space-before-tab,cr-at-eol diff --check

Fixes #1143
Related to #1047, #358