fix(model): wire openai_compatible configuration across trainer, eval, and config pipeline

[RohithPariki]

Description

This PR connects the generic openai_compatible backend across the configuration, trainer, and evaluation pipelines so that models served via OpenAI-compatible endpoints (such as DeepSeek, Groq, Together AI, vLLM, Ollama, LiteLLM, or local servers) can be fully configured via YAML configs, --cfg-options, or CLI flags and properly initialized during training and evaluation runs.

Key Changes

1. Config pipeline (skillopt/config.py): Added all 18 model.openai_compatible_* mappings to _FLATTEN_MAP so YAML configs retain general and per-role (optimizer_ / target_) settings during flattening.
2. Backend runtime configuration (skillopt/model/openai_compatible_backend.py & skillopt/model/__init__.py): Extended configure_openai_compatible to support per-role overrides for temperature, timeout_seconds, and max_tokens across OPTIMIZER_CONFIG and TARGET_CONFIG.
3. Trainer initialization (skillopt/engine/trainer.py): Added configure_openai_compatible(...) call during ReflACTTrainer.train() initialization so parameters configured via YAML or CLI take effect during training and episode rollouts.
4. Standalone evaluation harness (scripts/eval_only.py): Added "openai_compatible", "qwen", and "qwen_chat" to --backend choices, added CLI flags, mapped them to structured keys in load_config, and invoked configure_openai_compatible in main().
5. Training CLI (scripts/train.py): Added CLI arguments for openai_compatible_*, mapped them in _LEGACY_TO_STRUCTURED, and added environment variable guidance (OPENAI_COMPATIBLE_API_KEY) for secure credential passing.
6. Unit Tests (tests/test_openai_compatible_config.py & tests/test_openai_compatible_backend.py): Added comprehensive test coverage for YAML config flattening, CLI overrides, role separation, trainer initialization, credential warnings, and eval_only script parsing.

Verification

• Ran test suite across config and backend modules (pytest tests/test_openai_compatible_config.py tests/test_openai_compatible_backend.py tests/test_azure_openai_compat.py tests/test_minimax_backend.py tests/test_minimax_region.py tests/test_qwen_backend.py tests/test_role_backend_resolution.py tests/test_codex_config_aliases.py tests/test_retired_cli_options.py tests/test_env_section_survives_dedup.py). All 175 tests passed.

[Yifan Yang (Yif-Yang)]

Thanks — the new OpenAI-compatible fields flatten and reach the runtime configurator, but current head still has several entry-point gaps:

1. scripts/train.py does not include openai_compatible in the --backend choices, so --backend openai_compatible exits with argparse status 2 even though the related flags were added.
2. scripts/eval_only.py has no qwen_chat routing branch. Because both qwen and qwen_chat normalize to qwen_chat, --backend qwen currently falls through to the generic branch and selects openai_chat for both roles; the target therefore never uses Qwen. Eval-only also lacks the Qwen model-default normalization already present in train.py, so an inherited gpt-5.5 can remain on a Qwen role.
3. Eval-only parses and maps all optimizer_qwen_chat_* options, but its configure_qwen_chat(...) call forwards only shared and target values, so every optimizer-specific Qwen setting is silently ignored.
4. OpenAI-compatible role selection is correct, but model fallback is not handled in either train or eval-only. With the shipped base config, selecting openai_compatible leaves both role models at the inherited Azure gpt-5.5; with no openai_compatible_model override, that becomes the actual compatible wire model instead of the backend's declared gpt-4o-mini fallback.

Please wire these paths consistently while preserving explicit per-role model overrides, and add regressions for train CLI backend selection, eval-only qwen/qwen_chat role and model resolution, optimizer Qwen kwargs, and OpenAI-compatible fallback/explicit-model precedence. The newly added compatible tests currently all pass but do not exercise these cases.

[RohithPariki]

Thanks Yifan Yang (@Yif-Yang) for the detailed review! I have addressed all 4 points in commit 451f2e8:

1. train.py --backend choices: Added "openai_compatible" to the argument choices in scripts/train.py.
2. eval_only.py Qwen routing & model normalization: Added the qwen_chat role routing branch (optimizer_backend = "openai_chat", target_backend = "qwen_chat") and model-default normalization for qwen_chat roles to resolve the target model to Qwen/Qwen3.5-4B.
3. eval_only.py optimizer Qwen kwargs: Updated the configure_qwen_chat(...) invocation to forward all optimizer_qwen_chat_* options.
4. OpenAI-compatible model fallback & precedence: Added normalization in both train.py and eval_only.py so that unoverridden base gpt-5.5 sentinels fall back to --openai_compatible_model / per-role compatible models or the declared default gpt-4o-mini, while strictly preserving explicit per-role model overrides (--optimizer_model / --target_model).
5. Regressions: Added unit tests in tests/test_openai_compatible_config.py covering train CLI backend selection, eval-only Qwen role and model resolution, optimizer Qwen kwargs forwarding, and OpenAI-compatible fallback / explicit-model precedence.