data: recover 846 GSMArena budget phones via SoC-matcher fix

[Seungpyo1007]

What & why

Follow-up to PR #98. While importing cigarplug/scrape-gsma I found a bug in my SoC-matcher's die-number extraction (it stripped hyphens, breaking the \b word boundary so mediatek-mt6582 never matched the chipset Mediatek MT6582). Fixing it recovers 846 budget/mid-range phones — mostly MediaTek MT6xxx devices — that link to already-existing SoCs (no new SoC records needed).

• Added: 846 smartphones (BLU, Allview, Verykool, Lenovo, Micromax, Alcatel, ZTE, Panasonic, Gionee, and other budget brands)
• Modified/Deleted: 0
• All verified: false, integer ram_gb, valid soc FK to the existing catalog.

Commits (each Refs #1): 4 brand-group smartphone commits + surgical dump refresh.

Closes #1

Source

GSMArena via cigarplug/scrape-gsma, *-scrapegsma-<row> slugs; rows previously skipped only because the die-token matcher missed their SoC.

Dump

Regenerated via TechEngine app.dump at pinned submodule c0928f9 (scoring v2.0.0). Only the 846 new records' detail+score files, the smartphones index (92,497→93,343) and the manifest are committed — no churn on unchanged records.

Checklist

• python -m app.validate — ✅ passed
• python TechEngine/integrity_check.py data --strict — ✅ no hard anomalies
• git diff --check — ✅ clean
• Correct paths, kebab-case unique slugs, source_urls present

[Seungpyo1007]

🤖 Review — PASS

Check	Command	Result
Schema/self-check	python -m app.validate	✅ passed (full dataset)
Integrity gate	python TechEngine/integrity_check.py data --strict	✅ no hard anomalies
Whitespace	git diff --check origin/main..HEAD	✅ none
Engine dump	app.dump @ c0928f9	✅ 105,671 records, no serializer errors

Data-change summary

• Added: 846 smartphones · Modified/Deleted: 0
• All verified:false, integer ram_gb, valid soc FK (all 846 verified against the SoC catalog: 0 invalid).
• Root cause fixed: SoC die-token matcher stripped hyphens → \b boundary broke → mediatek-mt6582 never matched Mediatek MT6582. Now die numbers (MT6xxx, SM8xxx, MSM8xxx, SDM6xx) extract correctly.

Example records

• data/smartphone/acer/2012/liquid-gallant-duo/acer-liquid-gallant-duo-scrapegsma-1759.json (soc mediatek-mt6575)
• BLU/Allview/Verykool budget devices on MediaTek MT6580/MT6582/MT6737.

Site build

• No site/src changes. Surgical dump refresh (846 new records + smartphones index + manifest only).

[Seungpyo1007]

📊 Dataset stats (post-merge projection)

Metric	Count
Total records	105,671
verified: true	10,028
verified: false	95,637
missing verified	6
Verified ratio	9.49%

Warning

Verified ratio 9.49% — expected for the bulk-import campaign; promotion happens later via TechEngine verify-network. The 6 missing-verified records pre-date this PR.

Session running total (cigarplug/scrape-gsma fully mined): PR #97 (373) + PR #98 (2,498) + this (846) = 3,717 net-new records. cigarplug is now exhausted; further phone growth needs new datasets or a new category (see #1).