Not a bad idea, but while you've changed the format in how many of the games of chance are played, it's still essentially the same rules as many standard gambling games. The firework game, for example, is slots with a different skin. You thought way outside of the box with a lot of ATITD's games, Seven Blades for example, it stands to reason that if you really put your mind to it you could create some gambling games for this that would be very unique.
Actually, Sevenblade is Liar's Poker, re-skinned.
But you're right - there are a limited number of pure games of chance, unless you introduce decision making or skill-based PvP.
There will be games with both decision making and PvP, but there's a nasty little game design secret that tells me that these will account for a tiny fraction of the games played in Dragon's Tale. The reasons are related:
Skill-based PvP games are fun for the people that tend to win, and frustrating for the people that tend to lose. This was overwhelmingly obvious in Tales 1 and 2: We had an entire Discipline (1/7th of the game) that revolved around 2-player abstract games. It was a hugely popular discipline at first, but those people that tended to lose most of the time would get frustrated and quit the game, leaving a smaller, smarter population of Conflict Disciples. Then the less-smart half of those that remained would get frustrated and leave, etc., until we ended up with about a half dozen total geniuses. (Two of them were in fact Math PhDs.)
Decision-Making games have a related problem: In order to set rates of return, I need to be sure I know the optimal play strategy. If I just sort of "wing it", then it's a matter of time before someone discovers a strategy with a 100%+ return and writes a macro to bankrupt the casino.
An alternative approach is to have dynamic payback rates (which is the method that I've decided to use.) A daemon will watch overall player returns and adjust either the cost to play, or the payoff table. With this method, it's possible for the smartest players to play with 100%+ return, but still have the house profit overall. Returns for typical players will then tend to be quite a bit lower than all the other games in Dragon's Tale, because smart players will play very heavily. (Or even write macros to play for them, quickly and perfectly.)
So I think that both types of games will have very few players, and that fun & easy games of pure chance will dominate. But we'll see - I like doing experiments