I would do a little experimentation. Try it both ways, and see which gives the best combination of idle and part-power performance and economy. With three carburetors all linked together, each carburetor will only be opening about one-third as much as a single carburetor would be to supply the same amount of power. Thus, the "metered vacuum" from any one of the carburetors in a triple set-up will be different from what would be produced from a single carburetor manifold, as the metered-vacuum port is usually just slightly above the fully-closed throttle plate. Where is it on a Stromberg BX-35?
Get a couple of vacuum gauges, connect one to the manifold and the other one to one of the "metered" ports on the carburetors, and see just how the vacuum signals differ during the different modes of operation - idle, off-idle, low speed cruise, high speed cruise, wide-open, etc. The primary reason for running for running progressive throttle linkage is to prevent the linked-together carburetors from operating in the off-idle mode (kind of a no-man’s land for the carburetor) during low speed cruise. If you have a wideband air-fuel meter, you have the makings of a real science fair project. I think that Frank may have done some of this.
Back in the late fifties-early sixties, I ran a 235 with two Stromberg BXOV-2 carbs, and a couple of 261s with both the two Strombergs single barrels and with three 97s. With the 97s I had no choice but to use manifold vacuum, and I found that manifold vacuum seemed to provide better overall satisfaction than carburetor vacuum to the distributor with the dual BXOV-2s. Perhaps the results would have been different if I had modified the centrifugal advance.
If I recall correctly, some GM cars from fifty years ago also sent manifold vacuum to to the distributor at idle. Some cars also have dual diphragm arrangements, with an additional vacuum retard function.
