spec-guards: 5 more stale seals after the 2026-10-07 merge wave (knowledge_graph, prims_mst, graph_bfs, formats_catalog, workflow)

Problem

After #7360, spec-guards still fails on seals left stale by the 2026-10-07 07:24-07:40Z merge wave:

Spec	Merge	Kind
`specs/graph/knowledge_graph.t27`	#6872	spec changed, not resealed
`specs/tri/graph/prims_mst.t27`	#6876	spec changed, not resealed
`specs/tri/graph/graph_bfs.t27`	#7065	spec changed, not resealed
`specs/numeric/formats_catalog.t27`	#6930	seal minted before the final spec edit
`specs/tri/pipeline/workflow.t27`	#6994 (gen-rust enum order)	gen-drift, spec unchanged

workflow: 3/3 tests pass (same as the old seal).
formats_catalog: 0/0 (a data catalog with no tests, same as before).
knowledge_graph, graph_bfs: tests BLOCKED at Zig compile, as their old seals already recorded (same error kinds).
prims_mst: tests BLOCKED (spec.zig:95:49: expected '}', found ']'). The old seal predates test reporting. The body Implement the 1 empty function body in specs/tri/graph/prims_mst.t27 #6876 added is not a working MST: allocator : "std.mem.Allocator" is a string type, mst returns void and drops its result, and both tests check only result != undefined. Needs a real rewrite (separate work).

Blocked is recorded as blocked, never as a pass.

Reseal the five specs; check that check_seal_currency.py and check_seal_coverage.py both exit 0 against current master.