feat/upgrade open ai model to 5.4mini

[kdh29]

The model has been changed to 5.4 mini with reasoning set to low. testing showed similar latency with the previous model. If reasoning proves to be completely unnecessary then it can be turned off by setting reasoning to off.

[kcarnold]

Typechecker fails in experiment:

$ npm run build
...
./app/api/chat/route.ts:12:35
Type error: Expected 1 arguments, but got 2.

  10 |
  11 |   const result = streamText({
> 12 |     model: openai('gpt-5.4-mini', { reasoningEffort: 'low' }),
     |                                   ^
  13 |     system: scenario.chat.systemPrompt,
  14 |     messages: convertToModelMessages(messages),
  15 |     maxOutputTokens: 300,
Next.js build worker exited with code: 1 and signal: null