hostbuf_file_reader_read failed with all mixed precision quantized video models on RTX 5090 Laptop

## System Information
- ComfyUI Version: 0.24.0
- comfy-aimdo Version: 0.4.8
- PyTorch Version: 2.12.0+cu130
- GPU: NVIDIA GeForce RTX 5090 Laptop GPU (24GB VRAM)
- OS: Windows 11
- Python: 3.13.12

## Problem
All mixed precision quantized video models fail with the same error:
RuntimeError: hostbuf_file_reader_read failed

## Affected Models
- LTX-2.3 22B Dev FP8
- Wan2.2 T2V (low noise and high noise)
- Wan2.1 I2V 480P

## Error Location
File: comfy_aimdo/host_buffer.py, line 71, in read_file_to_device

## Notes
- Image generation (Flux.1) works perfectly fine
- All video models that contain quantization metadata trigger this error
- The error occurs consistently after "Model Initializing" phase
- Downgrading comfy-aimdo to 0.4.8 did not fix the issue

Start by reproducing the failure with one of the listed mixed-precision video models and inspect comfy_aimdo/host_buffer.py around line 71 in read_file_to_device. Compare that path with the working Flux.1 image-generation case and the model initialization logs. Done means the affected video models initialize without the hostbuf_file_reader_read error on the reported RTX 5090 Laptop setup.

Written by the indexing model from the issue text.