When using v1 model in Windows build, libvmaf ERROR could not initialize feature extractor "Speed_chroma_feature_speed_chroma_uv_score" #1556

I'm using the binary file from https://github.com/Netflix/vmaf/releases/tag/v3.2.0. The parameter is like this:
['-r', sourceY4m, '-d', destY4m, '-o', vmafCsv, '--csv', '--threads', '12', '--model', 'path=./model/vmaf_v1.0.16/vmaf_v1.0.16_3d0h.json']
This works correctly.

Next, when I use parameter --model path=./model/vmaf_v1.0.16/vmaf_v1.0.16_3d0h.json, this fails with:

libvmaf ERROR could not initialize feature extractor "Speed_chroma_feature_speed_chroma_uv_score"
problem loading feature extractors from model: ./model/vmaf_v1.0.16/vmaf_v1.0.16_3d0h.json

I suggest the binary might be built incorrectly without this feature.

> vmaf --version
3f9e02af

It might be inconvienient for me to build the binary myself. How can this problem be solved? Thank you!

By the way, the reason why I want to use the new model is that I meet the abnormal result with the default model by mjpeg:

The brown line shows the result (score / video time) by libx265. The green line show the result by mjpeg. Even when I evaluate the poorest quality mjpeg video (Disgusting to the naked eye), the result almost stuck at 100.
I don't know if I was using it incorrectly.