Question: Why does `applyrule` fail to apply rules?

I would like to simulate the evolution of the so-called B, C, K, W system and SKI combinator calculus in Maple.
The rewrite rules of them are simple: 

However, if I try to evaluate the following example given in the Wikipedia article, 

Maple will only return an unchanged result: 

restart;
applyrule([cS(x::anything)(y::anything)(z::anything) = x(z)(y(z)), 
   cK(x::anything)(y::anything) = x, cI(x::anything) = x], 
  cS(cK(cS(cI)))(cS(cK(cK))(cI))(x)(y)); # Unable to reduce??? 
 = 
              cS(cK(cS(cI)))(cS(cK(cK))(cI))(x)(y)

I believe that this is not an outlier.
Here are two additional instances: 

> rls := [cS(x::anything)(y::anything)(z::anything) = x(z)(y(z)), cK(x::anything)(y::anything) = x]:
> map2(applyrule, rls, [cS(cS(cS)(cS))(cS)(cS(cS))(cK), cS(cS(cS))(cS)(cS)(cS)(cS(cS)(cK(cK)))]);
                       [cS(cS(cS)(cS))(cS)(cS(cS))(cK), cS(cS(cS))(cS)(cS)(cS)(cS(cS)(cK(cK)))]

>

So why can't `applyrule` apply rules as desired? Meanwhile, how to automatically and thoroughly (like :-eval['recurse'] or MmaTranslator:-Mma:-ReplaceRepeated) apply those transformation rules to 

1. cS(cK(cS(cI)))(cS(cK(cK))(cI))(x)(y), 
2. cS(cS(cS)(cS))(cS)(cS(cS))(cK), and
3. cS(cS(cS))(cS)(cS)(cS)(cS(cS)(cK(cK)))? 

I have read something like How to apply a recursive rule in an expression? - MaplePrimes, but they are not the same issue.