feat: Add native OpenAI (GPT) direct provider

[mukund-gohil-atos]

Summary

OpenAI GPT models work today via --provider openai-compatible by pointing --base-url
to https://api.openai.com/v1, but there is no named openai provider shorthand. This
causes confusion since users expect to type --provider openai the same way they can
type --provider anthropic.

Problem

• Users must know to set --base-url https://api.openai.com/v1 manually
• No clear provider-specific validation or defaults for OpenAI
• gpt-4o, o3, o1 models require max_completion_tokens instead of max_tokens
  — this is handled internally but never surfaced to users

Proposed Solution

Add an explicit openai provider option:

• codewiki config set --provider openai --api-key YOUR_OPENAI_KEY --main-model gpt-4o
• Auto-sets base_url to https://api.openai.com/v1
• Documents the max_completion_tokens behaviour for o-series models

Files Affected

• codewiki/src/be/llm_services.py — add openai provider routing
• codewiki/src/config.py — add openai to provider list with sensible defaults
• README.md — add OpenAI example

Acceptance Criteria

• --provider openai accepted, base URL auto-set
• Works with gpt-4o, o3, o1 models out of the box
• README updated