L4 dataflow: formal_in ports have no outgoing ddg; statement↔actual-port links absent (partial twin of the canpy port-layer gap)

[rahlk]

Twin of the codeanalyzer-python L4 port-layer audit (filed today; found during python-sdk#270 verb validation). On the cants L4 sample in python-sdk (tests/resources/cpg/ts-a4.json):

• stmt → formal_out already emitted (3/3 formal_out ports wired) — ahead of canpy here.
• formal_in ports: 2, 0 with outgoing ddg — parameter flow entering via param_in dead-ends at the port (the @entry → stmt var edges duplicate the role).
• No stmt → actual_in or actual_out → stmt edges observed — argument-binding and return-value-binding at call sites are missing, so cross-callable flows_to witnesses are port-to-port segments only.

Caveat: the committed sample is small (one call site); please verify the absence classes on a larger -a 4 corpus before scoping — the canpy audit used a 758-callable corpus and the pattern there was categorical (zero port-touching ddg edges of any class).

Missing edge checklist (same as canpy, minus the return side already done):

• def stmt → actual_in:k
• actual_out → use stmt
• formal_in:k → first-use stmts

Also worth checking whether TS call vertices sit on the CFG spine (canpy's are 100% cfg-orphaned).