AI responses now use predictable output and thinking budgets

Objo Studio AI assistant responses use a consistent output budget across providers and apply the thinking preference through provider-appropriate request translation. This makes response limits predictable while preserving each provider protocol.