We see from Turing’s On Permutations paper that massive amounts of non-electronic-computation based MC can be occurring, before 1954 (by definition since that’s when Turing died). Obviously, it could be occurring as early at 1941, or more likely 1946 when lots of Turing bombes are looking for “other purposes”. Remember, we don’t know when that manuscript was written (though I suspect its late 30s for reasons of Turing’s 1920s writing style, and its evolution over time).
While its true that by 1950s on Manchester computers Turing is modeling such processes as were being modeled in the mentioned the MANIAC computing efforts using his floating point registers, its also clear that he could have been doing much the same thing with high-speed bombes too, a decade earlier.
Turing would have been perfectly placed to have take 1944-era nuclear fission simulation work and apply bombes as calculating devices, calculating MC on graphs.