Fix CI test FP8 static backend matching failure

[lvliang-intel]

Description

Problem
test_fp8_static and test_fp8_static_rtn fail during model loading with:
ValueError: No compatible backend found for layer model.decoder.layers.1.self_attn.v_proj
with config QuantizationScheme(..., transform_config={})

Root Cause
When a quantized FP8 static model is saved and reloaded, the transform_config field undergoes a None → {} drift:

Preset definition (FP8_STATIC): transform_config=None (dataclass default)
Export/save: BaseCompressor initializes transform_config to {}, which gets serialized into the quantization config JSON
Load/inference: JSON deserializes it back as {}, producing QuantizationScheme(..., transform_config={})
Backend matching: fp8_static_scheme_checker compares the loaded config against FP8_STATIC using ==, which fails because {} != None

Fix
Updated QuantizationScheme.eq to treat None and empty dict {} as semantically equivalent for dict-typed fields like transform_config. Both represent "no transform config", so they should compare as equal.

Type of Change

• Bug fix
• New feature
• Documentation update
• Performance improvement
• Code refactoring
• Other (please specify):

Related Issues

Fixes or relates to #

Checklist Before Submitting

• My code has been tested locally.
• Documentation has been updated as needed.
• New or updated tests are included where applicable.

[Copilot]

Pull request overview

Fixes FP8 static backend matching during model reload by making QuantizationScheme equality tolerant to None vs {} drift in dict-typed fields (notably transform_config), preventing false “no compatible backend” errors in FP8 static inference.

Changes:

• Updates QuantizationScheme.__eq__ to treat None and empty dict {} as equivalent for dict-valued fields.
• Prevents FP8 static backend selection from failing when reloaded configs deserialize transform_config as {}.