Allow `-max-ai-credits=0` when using local models

### Describe the feature or problem you'd like to solve

I'd like to be able to use local models with `/remote` without the possibility of accidentally using credits in that session

### Proposed solution

Remote the validation on `-max-ai-credits` that requires that it is >= 30.

### Example prompts or workflows

_No response_

### Additional context

The limit of 30 seems totally arbitrary, and when using a local model it doesn't mean anything. When testing with local models, it would be nice to know for sure that we can't accidentally use credits.

首先定位 `-max-ai-credits` 选项的验证逻辑，以及在使用 `/remote` 时识别本地模型的代码。确认当前最小值为 30，然后验证本地模型工作流接受值 0，并确保会话不会意外使用额度。

由索引模型根据 Issue 内容生成。