fix: address Bugbot findings from PR #22 (target_size order + Makefile/CI lint parity)

[aptracebloc]

Summary

Addresses the two Cursor Bugbot findings raised on the develop → main sync PR #22. Per our flow, the fixes land on develop here; #22 then picks them up automatically and the Bugbot threads resolve.

1. target_size emitted as [width, height] — schema/ingestor want [height, width] (Medium)

The embedded ingest.v1.json documents target_size as [height, width] and the ingestor reads it that way, but buildImage wrote [width, height] (the order from --target-size WxH / DetectImageSize). Square datasets were unaffected — which is why it slipped through — but non-square datasets would pass local validation yet mis-validate resolution in the ingestor.

• Swapped both emit sites (keypoint top-level + image spec.file_options) to [height, width]; kept the user-facing --target-size WxH flag and the auto-detect message unchanged.
• Added TestBuild_TargetSize_EmittedHeightWidth — a non-square (640×480 → [480, 640]) regression test, since square sizes can't catch a swap.

2. make ci lint diverged from CI lint (Medium)

make lint ran golangci-lint, while CI's lint job runs errcheck + gofmt -s + ineffassign + misspell (golangci-lint-action is disabled pending #6). The Makefile exists specifically to keep local == CI.

• make lint (used by make ci) now runs the same standalone tools as CI (gofmt -s is fmt-check, go vet is vet), restoring the "make ci green ⟹ CI green" invariant.
• golangci-lint moved to make lint-full for an optional richer pass; re-unifying on it is tracked in Re-enable staticcheck + unused linters once we have a strategy for the k8s.io dep tree #6.

Verification

make ci green end-to-end: go vet, go test -race, the new standalone lint, fmt-check, and schema-check (no drift).

🤖 Generated with Claude Code

---

Note

Medium Risk
Changes dataset push spec output for non-square resolutions (fixes ingestor mis-validation) and redefines local lint to match CI; both affect correctness and developer workflow but not auth or data paths directly.

Overview
Ingest spec target_size order: buildImage now emits [height, width] (schema/ingestor order) while SpecArgs.TargetSize stays [W, H] from --target-size / auto-detect. Both emit paths swap on output: keypoint top-level target_size and image spec.file_options.target_size. Comments on TargetSize document the storage vs emit convention.

Regression test: TestBuild_TargetSize_EmittedHeightWidth uses 640×480 so stored [640, 480] must become [480, 640] for image_classification and "and" keypoint_detection (square sizes would not catch a width/height swap).

Makefile / CI parity: make lint (part of make ci) runs the same standalone tools as .github/workflows/build.yml — errcheck, ineffassign, and misspell via go run — instead of golangci-lint. make lint-full keeps the optional golangci-lint run pass until golangci-lint-action is re-enabled (#6).

^{Reviewed by Cursor Bugbot for commit 34e1755. Bugbot is set up for automated code reviews on this repo. Configure here.}

[LukasWodka]

👋 Heads-up — Code review queue is at 13 / 8

Above the WIP limit. The team convention is to review existing PRs before opening new work.

Open PRs currently in Code review (oldest first):

• averaging-service#98 — ci: add fr-gate caller workflow · author: @LukasWodka · no reviewer assigned
• cli#22 — Sync develop → main for v0.1.0-alpha · author: @aptracebloc · no reviewer assigned
• cli#24 — docs: refresh README for the v0.1.0-alpha release · author: @saadqbal · no reviewer assigned
• client#191 — fix(#190): fail image-refresh loudly when the ingestor (ghcr) digest can't resolve · author: @saadqbal · no reviewer assigned
• client-runtime#75 — dev to staging · author: @shujaatTracebloc · reviewer: @saadqbal
• data-ingestors#146 — test(schema): lock category enum to the engine's TaskCategory (anti-drift) · author: @LukasWodka · reviewer: @saadqbal
• design-system#19 — fix: un-track coverage/ and node_modules/ from git · author: @LukasWodka · no reviewer assigned
• design-system#22 — ci: add Vitest test workflow · author: @LukasWodka · reviewer: @aptracebloc
• design-system#23 — ci: add fr-gate caller workflow · author: @LukasWodka · no reviewer assigned
• docs#46 — docs: make declarative-ingest staging self-contained (data-ingestors#131 B/C) · author: @divyasinghds · no reviewer assigned

Pull from review before opening new work. (This is a nudge from the kanban WIP check, not a block.)