I guess we should really read the articles to find out what they classify as "optimal strategy".
I could believe that there exist strategies that would be effective against certain players/strategies than their "optimal strategy" (by "more effective" I suppose I mean "would beat them faster"), but that does not mean that their "optimal strategy" against any oponent does not exist.
On what basis do you make the claim "The programmers assume an optimal strategy for poker, but there is no such optimum strategy as such."?
Tick-tack-toe, chess, and go all clearly are not "games of skill" if you have the knowledge of all possible board configurations and their interconnections. For chess and go it is not clear when (if ever) our storage and calculating abilities will completely be able to map out the space, but the space is defined. Since there is no chance element involved it is easier to see how knowledge of the space translates into the optimal moves.
For games like poker, craps, and rock-paper-sissors(-lizard-Spock?), it does become a bit more difficult to define "optimal" since each hand or round is non-deterministic, so there is no possiblity of guaranteeing a win in any particular round. In RPS, there is a strategy that is guaranteed to win at least 50% of the time, against any possible other strategy. Even knowing that your opponent is using this strategy does not give you enough information to consistently beat it. Why do you think that such a strategy does not exist for poker?
Is there a name for one-card poker? Each player gets one card, you bet/call/raise/fold and you are done. Has that been "solved"? A quick web search turns up a three-card-deck variant: https://en.wikipedia.org/wiki/... which has been analyzed back in 1950. Is there some qualitative difference to suggest that more complex forms of poker cannot be analyzed in a similar way?