feat(chat): add verbosity parameter support

[pwerry]

Q	A
Bug fix?	no
New feature?	yes
BC breaks?	no
Related Issue	N/A

Describe your change

This PR adds support for the new verbosity parameter in ChatCompletionRequest, which was introduced with GPT-5 and allows controlling response length and detail.

Implementation details:

• Created Verbosity value class with predefined Low, Medium, High options
• Added verbosity parameter to ChatCompletionRequest and ChatCompletionRequestBuilder
• Follows the same pattern as existing reasoningEffort parameter
• Added comprehensive tests and updated CHANGELOG.md

Usage:

val request = chatCompletionRequest {
    model = ModelId("gpt-4o")
    verbosity = Verbosity.Low  // or Medium, High
    messages { /* ... */ }
}

What problem is this fixing?

The OpenAI API recently introduced a verbosity parameter for Chat Completions and Responses API endpoints. This parameter allows developers to control how concise or verbose the model's responses are:

• low: Terse, concise responses
• medium: Balanced responses (default)
• high: Detailed, comprehensive responses

Without this parameter, developers cannot take advantage of this new API feature to optimize response length for their specific use cases.