Train an issuer baseline from known-good PDFs

multipart/form-data. Fields: issuer (slug, ^[a-z0-9][a-z0-9-]{0,63}$), label (optional human name), and up to 25 file parts, each within the deployment's upload cap.

Send issuer and label BEFORE the file parts: an invalid slug is then rejected before a single document is parsed.

Each file is parsed in memory, reduced to its structural profile, and discarded. Training documents are never written to disk. Files that are not PDFs, or that cannot be parsed, are skipped and reported in skipped; they do not fail the request unless none is usable.

Training an issuer that already has a profile MERGES into it, so a baseline can be grown in batches. Training does not consume inspection quota.

Team plan and up. The built-in public baselines work on every tier; training private ones answers 403 plan_required below Team.