Support GPT-5 models token parameter (`max_completion_tokens`)

[calvesrodrig]

Is your feature request related to a problem? Please describe.

When using new OpenAI GPT-5 models in the Obsidian Copilot plugin, requests fail with the following error:

Error: Model request failed: 400
Unsupported parameter: 'max_tokens' is not supported with this model.
Use 'max_completion_tokens' instead.

This happens because GPT-5 models no longer support the legacy max_tokens parameter, but the plugin still sends it in the request payload.

---

Describe the solution you'd like

Update the OpenAI request handling so that:

• GPT-5 models use max_completion_tokens
• Legacy and older models continue using max_tokens, when required

Ideally, this should be handled automatically based on the selected model, without requiring manual configuration from the user.

---

Describe alternatives you've considered

• Manually modifying the plugin source code to replace max_tokens with max_completion_tokens
• Downgrading to older OpenAI models that still support max_tokens

Both alternatives are not ideal for long-term maintenance or for users who want access to newer models.

---

Additional context

According to OpenAI’s updated API behavior, GPT-5 models require max_completion_tokens instead of max_tokens.

This change currently prevents GPT-5 models from being used at all in the Copilot plugin.