Two sources, because the two figures are not the same kind of fact. Weight, price and introduction year belong to the design wherever it turns up, so they go in one table keyed by designation - scripts/unit-stats.py, collected from what real MegaMek already reported about the dailies. Battle value belongs to the machine *and its crew*, because BV 2.0 multiplies by gunnery and piloting, so it rides on the scenario: `mms index` now writes one per unit into daily.json. Stinger STG-3R is the proof, at 359 in one daily and 323 in another. The report lists a faction's units in file order and the card orders them across the board, so the two are matched by the position each unit started in. By name would be wrong the moment a force fields two of the same design with different crews, which is a normal thing for a scenario to do. Coverage is partial and stays visible as partial: nine of the thirty-four machines the hand-written catalog fields have never appeared in a daily, and `statsFor` returns null for them rather than a zero.