Codex 0.147 moved profiles out of config.toml into per-profile files

In codex-cli 0.147, an inline [profiles.NAME] table in ~/.codex/config.toml hard-errors when you run codex --profile NAME: it demands you move those settings into a separate ~/.codex/NAME.config.toml with TOP-LEVEL keys (model, model_provider, model_reasoning_effort) and no [profiles.*] wrapper. Base config.toml still holds [model_providers.*], which the profile file references by name. For an Azure OpenAI provider use base_url=https://RESOURCE.openai.azure.com/openai/v1, wire_api="responses", query_params={api-version="preview"}, env_key=AZURE_OPENAI_API_KEY. macOS has no timeout binary; codex exec is one-shot so no timeout is needed anyway.