perf: Run quantization on CPU, remove ZeroGPU timeout issue

- Remove @spaces.GPU decorator - simple mode is pure tensor math
- Remove low_memory flag - 192GB RAM available
- Remove unused spaces import

app.py

@@ -7,7 +7,6 @@ FP8/INT8 formats for ComfyUI inference, with HuggingFace Hub integration.
 import os
 import tempfile
 import gradio as gr
-import spaces  # ZeroGPU
 from huggingface_hub import hf_hub_download, HfApi, create_commit, CommitOperationAdd
 
 from convert_to_quant import convert, ConversionConfig
@@ -147,8 +146,6 @@ def upload_model_as_pr(
     except Exception as e:
         return f"❌ Upload failed: {str(e)}"
 
-
-@spaces.GPU
 def quantize_model(
     source_repo: str,
     file_path: str,
@@ -217,7 +214,6 @@ def quantize_model(
         save_quant_metadata=True,
         simple=True,
         verbose="VERBOSE",
-        low_memory=True,
         scaling_mode=format_config.get("scaling_mode") or "tensor",
         block_size=format_config.get("block_size"),
         filter_flags=filter_flags,