feat: Add mapping for max_tokens for groq and cerebras

[tibisabau]

Description:

• Fixed max_completion_tokens parameter mapping for GROQ provider to map to max_tokens instead of max_completion_tokens
• Fixed max_completion_tokens parameter mapping for Cerebras provider to map to max_tokens instead of max_completion_tokens
• Aligned GROQ and Cerebras with other providers (Deepinfra, Fireworks, Ollama, Oracle, etc.) that already map max_completion_tokens to max_tokens
• This ensures that when users send max_completion_tokens, it correctly maps to the provider's max_tokens parameter

Closes #1341.

The code in this pull request was generated by GitHub Copilot with the Claude Sonnet 4.5 model.

Tests Run/Test cases added:

• Build verification: npm run build passes successfully
• Code formatting: npm run format:check passes
• Pre-push validation: npm run pre-push completes successfully
• Unit tests: Created comprehensive test suite for max_completion_tokens mapping
  • tests/unit/providers/max-completion-tokens.test.ts - 11 tests passed
  • tests/unit/providers/parameter-transformation.test.ts - 12 tests passed
  • Total: 23 tests passed, 0 failed
• Configuration verification: Both GROQ and Cerebras correctly map max_completion_tokens to max_tokens

Type of Change:

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Documentation update
• Refactoring (no functional changes)