Missing --label-column on a tabular task gives a cryptic raw-schema error instead of naming the flag

[LukasWodka]

Found during functional review of the data-ingest flow on develop (2026-07-10).

What

Omitting --label-column for a label-requiring tabular task in non-interactive mode produces a raw JSON-schema dump that never mentions --label-column:

$ tracebloc data ingest ./data.csv --task tabular_classification --name t --dry-run --no-input
synthesized spec failed schema validation (2 issues):
 label: got string, want object
 label: minLength: got 0, want 1
Error: synthesized spec failed schema validation; check the flag values above

(exit 2)

Contrast: a wrong --label-column value gets a friendly guard — "Pass --label-column with one of the existing columns …" (internal/push/preflight.go:~148, CheckLabelColumn).

Why

Ordering: schema validation (internal/cli/data.go, step 4) runs before runLocalPreflight (which holds CheckLabelColumn), so an empty label="" trips the ingest.v1.json label oneOf(string|object) before the friendly guard fires. Generalizes to tabular_regression / time_series_forecasting (label: got object, want string), all exit 2, none naming the flag.

Scope

Non-interactive / CI only — interactive mode prompts for the label (internal/cli/interactive.go:299/376). Pre-existing (predates #210). Correct exit code (2); message quality only.

Fix direction

Detect the empty-required-label case before raw schema validation (or reuse CheckLabelColumn's message) so the error names --label-column and lists the candidate columns.

Severity: LOW. Verified on the real develop binary.

[aptracebloc]

Implemented and verified — closing.

Fixed by #233 (merged to develop 2026-07-13; body declares Closes #214, but the non-default merge target meant it didn't auto-close). Verified on origin/develop.

A targeted missing-label pre-check (internal/cli/data.go, step "3b") now runs before the raw schema validation (step 4) that produced the cryptic label: got string, want object dump:

if push.IsTabular(a.Spec.Category) && a.Spec.LabelColumn == "" {
 "this task needs a label column, but --label-column wasn't set —
 pass --label-column with the name of the target column in your data CSV
 (columns: <inferred CSV columns>)" // exit 2
}

Matches the fix direction exactly:

• Names --label-column and lists the inferred candidate columns.
• Intercepts only the missing case — a present-but-wrong label still flows to runLocalPreflight's CheckLabelColumn, and every other schema error still reaches the normal dump (no over-interception).
• Covers the whole label-requiring tabular/time-series family via push.IsTabular (classification, regression, time_series_forecasting).
• Correct exit code (2) preserved.

Test: TestDataIngest_TabularMissingLabel_Friendly (internal/cli/data_test.go).