Fix default OpenAI model unit test

[timrogers]

Motivation

• The default OpenAI model constant was updated to gpt-5.4-mini but the unit test still asserted the old gpt-4o, causing test failures.

Description

• Rename the unit test to default_openai_model_is_gpt_5_4_mini and update its assertion to assert_eq!(OPENAI_MODEL, "gpt-5.4-mini"); in src/llm.rs.

Testing

• Ran rustup run 1.94.0 cargo test --locked --all-features default_openai_model_is_gpt_5_4_mini (with COPILOT_SKIP_CLI_DOWNLOAD=1) which passed, and verified the full test suite with NO_PROXY=localhost,127.0.0.1,::1 COPILOT_SKIP_CLI_DOWNLOAD=1 rustup run 1.94.0 cargo test --locked --all-features which passed (59 passed; 0 failed), and confirmed NO_PROXY=localhost,127.0.0.1,::1 COPILOT_SKIP_CLI_DOWNLOAD=1 rustup run 1.94.0 cargo build --release --locked succeeded.

---

Codex Task