I think McKinley would win much more Western states since Cleveland and Bryan will split the vote that Bryan got there in RL and thus McKinley would win in many of them due to the divided Democratic vote. McKinley even won OR, ND, and almost SD when the Democrats were united in 1896 in RL. I agree that McKinley would win, though. Also, I think McKinley would easily win KY since he won it against Bryan in RL and possibly several other Southern states.
The West was fairly solid Free Silver, which was why the Populists did so well there in 1892 despite being a third party with no hopes of actually winning. With both McKinley and Cleveland running, the gold vote would be split, allowing the Populists to win some Western states that McKinley won in real life.
However, in the South, the Populist's appeal was less about Free Silver and more about providing an alternative to the Bourbon Democrats that the Republicans hadn't. If the Populists run a candidate of their own, many of McKinley's real life Southern votes would have gone to Bryan.
Finally, how do you make a three way map of 1896 when there are only 2 colors available for that year?
As the map below shows, there are five party colors available each year, but the evcalc page only provides easy access to however many actually had a chance that year. If you take a look the map url, you'll see that for each state there is a component, of the form:
<postalcode>=<colorcode>;<electoralvotes>;<decile><postalcode> is the postal code of the state
<colorcode> 1=red, 2= blue, 3=green, 4=yellow, 5=orange (also 0=gray, but then the decile doesn't matter)
<electoralvotes> However many electoral votes you want, or even text as I do with Texas below
<decile> Any number from 2 to 9 to indicate the voting strength was from 20 to 90%
However, I did it more simply than that. I used the 1892 evcalc as it had all three colors I wanted, then changed
year=1892 to
year=1896 and inserted
&UT=3;3;7 in the appropriate place for Utah, altho the order doesn't matter.
