Add Kimi K2 model and better support

[mrubens]

Adds the new Kimi K2 model to Groq, along with better support for models like this one where the max output tokens is the same as the context window.

---

Important

Add moonshotai/kimi-k2-instruct model and improve token handling for models with equal max tokens and context window.

• Models:
  • Add moonshotai/kimi-k2-instruct to groq.ts with 131072 max tokens and context window.
• Token Handling:
  • Update getModelMaxOutputTokens in api.ts to handle cases where maxTokens equals contextWindow by defaulting to 20% of context window.
  • Use getModelMaxOutputTokens in Task.ts to determine maxTokens for models.

^{This description was created by for d28587b. You can customize this summary. It will automatically update as commits are pushed.}

[delve-auditor]

✅ No security or compliance issues detected. Reviewed everything up to d28587b.

Security Overview

• 🔎 Scanned files: 3 changed file(s)

Detected Code Changes

Change Type	Relevant files
Enhancement	► groq.ts     Add Kimi K2 model configuration ► Task.ts     Update token handling logic ► api.ts     Refine model max output tokens calculation

Reply to this PR with @delve-auditor followed by a description of what change you want and we'll auto-submit a change to this PR to implement it.