[Qwen/qwen3.5-9b] Reasoning setting 'medium' not supported - fallback to 'on'

Nobody has claimed this yet.

## Issue Description

A warning message appears in the developer console when using the qwen/qwen3.5-9b model via LLM Studio:

```log
2026-07-11 10:25:32 [WARN] Reasoning setting 'medium' is not supported by model 'qwen/qwen3.5-9b'. Supported settings: 'on', 'off'. Falling back to reasoning setting 'on'.
```

## Details

- **Model**: qwen/qwen3.5-9b
- **Error**: The UI allows selecting 'medium' as a reasoning setting, but this model only supports 'on' and 'off'
- **Impact**: When a user selects 'medium', the system falls back to 'on' automatically

## Expected Behavior

The reasoning dropdown should either:
1. Only show 'on' and 'off' for this specific model
2. Show a clearer error message instead of silently falling back
3. Validate the selected setting before applying it

This creates confusion when users see their selection not being honored, as they might think something else is wrong with their setup.

The issue does not name a file or test. Start at the reasoning-setting handling behind the dropdown and the qwen/qwen3.5-9b model configuration; determine whether unsupported values should be filtered, rejected, or reported more clearly. Done means the selected setting is handled consistently without an unexplained fallback.

Written by the indexing model from the issue text.