fd.suggest_plan / fd.plan raise TypeError on duplicate column labels #459

Severity: S2 (crash on a supported input) — found by the production-readiness fuzz campaign (FDC-L3-014)
Where: src/freshdata/engine/context.py:258 — build_context

fd.suggest_plan / fd.plan raise on a frame with duplicate column labels, which fd.clean accepts (it deduplicates them to "x", "x_2"):

TypeError: cannot convert the series to <class 'int'>

from int(s.isna().sum()), because df[label] returns a DataFrame rather than a Series.

Fix: select positionally, as the other APIs in this family now do — or raise the documented ValueError("... requires unique column labels ...") that fd.infer_roles / fd.explain_clean raise, so the behaviour is at least consistent.

Related: #445/#446/#452 (same duplicate-label family, other APIs), #265, #231 (closed).