[Misc] improve error messages for quantization mismatches and netloader config validation (vllm-project#10222)

### What this PR does / why we need it?
This PR improves error messages for three user-facing scenarios
identified during code audit, replacing bare Python exceptions with
clear guidance:

**1. Unsupported ModelSlim quantization scheme (`modelslim_config.py`)**

When a model is quantized with a scheme that vLLM Ascend recognizes but
doesn't fully implement (e.g., `quant_type=W4A16` registered for `moe`
but not for `linear`), the error previously only stated the fact without
suggesting a resolution:

```text
Before: NotImplementedError: Currently, vLLM Ascend doesn't support quant_type=W4A16 for layer_type=linear.
After:  NotImplementedError: ... Please use a supported quantization format or load the model with its original float weights.
```

**2. NetLoader crashes with bare AttributeError on malformed config JSON
(`netloader.py`)**

When a user passes a non-dict JSON value (e.g., a string or array) to
`--model-loader-extra-config`, `config.get()` on a `str`/`list` raises a
bare `AttributeError`:

```text
Before: AttributeError: 'str' object has no attribute 'get'
After:  RuntimeError: NetLoader requires --model-loader-extra-config to be a JSON object.
```

The fix adds an `isinstance(extra, dict)` check, following the same
pattern already used by `rfork_loader.py:43-46`.

- Fixes # (internal code audit, no GitHub issue)

### Does this PR introduce _any_ user-facing change?

Yes — error messages for three scenarios are changed:

1. When loading a ModelSlim model with an unsupported quantization
scheme (e.g., W4A16 + linear layer), the `NotImplementedError` now
includes usage guidance.
2. When `--model-loader-extra-config` contains a non-dict JSON value for
NetLoader, a `RuntimeError` with format guidance replaces the bare
`AttributeError`.

- vLLM version: v0.22.1
- vLLM main:
vllm-project/vllm@967c5c3
---------
Signed-off-by: xuchi <xuchicolson@163.com>

Author: xuchi-0808

vllm_ascend/model_loader/netloader/netloader.py

@@ -62,6 +62,12 @@ def __init__(self, load_config: LoadConfig):
 
         # Try to read config file at first
         extra = load_config.model_loader_extra_config
+
+        if extra is not None and not isinstance(extra, dict):
+            err_msg = "NetLoader requires --model-loader-extra-config to be a JSON object."
+            logger.error(err_msg)
+            raise RuntimeError(err_msg)
+
         if extra and "CONFIG_FILE" in extra:
             try:
                 logger.info("Reading configs in file %s ...", load_config.model_loader_extra_config["CONFIG_FILE"])

vllm_ascend/quantization/modelslim_config.py

@@ -393,9 +393,13 @@ def create_scheme_for_layer(
     if scheme_cls is not None:
         return scheme_cls()
 
-    err_msg = f"Currently, vLLM Ascend doesn't support quant_type={quant_type} for layer_type={layer_type}."
-    logger.error(err_msg)
-    raise NotImplementedError(err_msg)
+    err_msg = (
+        "Currently, vLLM Ascend doesn't support quant_type=%s for layer_type=%s. "
+        "Please use a supported quantization format "
+        "or load the model with its original float weights."
+    )
+    logger.error(err_msg, quant_type, layer_type)
+    raise NotImplementedError(err_msg % (quant_type, layer_type))
 
 
 @register_quantization_config(ASCEND_QUANTIZATION_METHOD)