eval: add VisualEars golden test set results (3 datasets)

eval: add Common Voice 17 results + consistent normalization, lead with CER

fix: --ignore-requires-python for omnilingual-asr 0.2.0 on Python 3.12

fix: pin omnilingual-asr==0.2.0 in install

docs: Open-in-Colab badge + point to notebook.ipynb

docs: add Rade AI contact info

docs: document fp16 single-file + precise two-path usage

Update WER/CER with Persian-normalized (hazm) numbers: WER 10.8% / CER 3.4%

Update model card: real benchmarks (fp16/fp32), WER/CER, install recipe

Add model card