feat: add rahimdzx/AraCode-7B-GGUF Arabic coding model

[Rahimdzx]

Summary

• Adds rahimdzx/AraCode-7B-GGUF to the curated TARGET_MODELS list in scripts/scrape_hf_models.py
• Adds a hardcoded FALLBACK entry — GGUF-native repos have no safetensors metadata so the scraper returns None; the fallback guarantees the record is always emitted
• Regenerated data/hf_models.json and llmfit-core/data/hf_models.json via the scraper

Model details

Field	Value
HuggingFace ID	rahimdzx/AraCode-7B-GGUF
Architecture	qwen2
Parameters	7.6B
Format	GGUF
Quantization	Q4_K_M (~4.68 GB)
Context	32768 tokens
Use case	coding
Languages	Arabic + English
License	Apache-2.0

This is the first open-source Arabic-specialized code explanation and generation model.

Test plan

• scrape_hf_models.py runs without errors
• rahimdzx/AraCode-7B-GGUF present in data/hf_models.json with correct fields
• llmfit list | grep -i aracode returns the model
• llmfit info "AraCode-7B-GGUF" shows correct params and quantization

[three-foxes-in-a-trenchcoat]

Overall: Good addition of AraCode-7B-GGUF as the first open-source Arabic-specialized coding model. The FALLBACK entry is a smart approach for GGUF-native repos without safetensors metadata. LGTM.

[three-foxes-in-a-trenchcoat]

Interesting data change — replacing tiny-random test models with real embedding models. Good direction, but this PR currently has merge conflicts (DIRTY state). Please rebase on latest main to resolve. The model additions themselves look fine once conflicts are cleared.

[AlexsJones]

Thanks for adding the AraCode-7B-GGUF model — the script changes look good and the GGUF fallback approach makes sense.

However, the regenerated data/hf_models.json and llmfit-core/data/hf_models.json have lost ~1,650 auto-discovered models (every entry with "_discovered": true is gone). The diff shows 1,904 models removed vs 263 added per file — that's a major data regression.

This likely happened because the scraper was run locally without the discovery pipeline that populates those entries.

Ask: Please don't regenerate the full JSON files. Instead, just add the new rahimdzx/AraCode-7B-GGUF entry to both JSON files directly (keeping all existing models intact). The script change to scrape_hf_models.py is fine as-is.

[three-foxes-in-a-trenchcoat]

Hey @Rahimdzx, please run cargo fmt to format the code.