The short answer to your analysis is this:

Players let the GM know they want to attempt a Risk and how they're going to approach it. The GM then decides if the Risk can be attempted and, if so, tells them what Trait/Skill combination to roll based on the situation. You can't just choose to use your highest Skill/Trait when it doesn't make sense and then improvise right away.

The long answer is, well, long. ;)

There's a handy sidebar on the upper-right corner of page 151 in the Core Rulebook that states GMs can deny players the possibility of success when it is "just silly or unrealistic." This applies to Advantages and applies to Risks as well.

Using the examples you've provided, here's how my GM would run them.

Fat Merchant: Under most circumstances Villanova would kill the merchant before the merchant can even lay a hand on Villanova. He's a Duelist Villain with a strength of at least 10. The Fat Merchant wouldn't even have a chance to harm him unless the villain was concentrating his attention elsewhere. Unless he surprises Villanova, the fat merchant in your example is dead before he can even do anything provided it's a one-on-one battle.

Here's how combat would likely go between these two:
Villanova has 6 Raises (he'd probably have even more but we'll go with just 6). Fat Merchant has 2 Raises.
Villanova slashes for 8 damage.
Villanova bashes for 1 damage.
Villanova slashes for 8 damage.
Villanova feints for 1 damage.
Villains go first on ties, so Villanova slashes for 9 damage. That's 27 Flesh Wounds. The merchant would be Helpless, but even if he could still continue, this is how the rest of the fight would probably go.
Fat merchant goes. He spends a raise but it does no damage because of Villanova's Bash from earlier.
Villanova bashes for 1 damage.
Fat merchant goes. He spends a raise but it does no damage because of Villanova's Bash from earlier.

Scholarship: I actually played in a convention game with John Wick and my character had zero ranks in Scholarship so he did not let me even attempt a roll to know a specific piece of information. Why? It didn't make sense for my character to know it.

The GM could simply say you wouldn't be able to roll for this. It happened to me exactly as described because my 2 Wits, 0 Scholarship character couldn't roll to know obscure knowledge. And this is coming from John Wick himself.

Vesten/Urchin: My GM would disallow the urchin from breaking down the door. They might be able to use Theft to pick the lock, though. There's more than one way to open a door, after all.

Hopefully this helps answer your question. Players will certainly be tempted to find ways to approach problems that utilize their best Trait/Skill combinations, but that doesn't mean the GM is going to let them if it doesn't make sense.

I'm glad Salty Dog provided his long answer. It saved me a lot of typing.

Assuming you haven't read it already, I serious suggest you read this essay by Wick. It really helped me to solidify the philosophy behind the mechanic. 

I would add that Consequences don't need to be consistent between the characters. In a Risk, the brawny Vesten is going to be able to buy off those consequences easier than the urchin, but in a dramatic scene I don't think the GM is off base asking for an additional raise or two from the urchin when he moves to break down the door (one to keep from alerting anyone nearby and another to get through the door without making repeated attempts). That's the thing about choices. Just because you CAN do something doesn't make it a good idea.

2 Wits and no scholarship? How is this even a risk? Why would you even have raises?

You take the analysis wrong way around. If you already have spent all raises, you have run out of time. Thus Knacks and Attributes does matter. You just make the normal mistake of normal RPG gamer, which assumes a check is for determining whether you succeed or not. THis is not case in 7th sean. In 7th sea Raises give out the information how well you succeed, and how much other things you can do besides succeeding.

THus in 7th sea, more skilled character can do more in the same scene turn than less skilled people. Roll gives out the random elements in the scene hindering the action you are doing, in addition to the scene derived hinderances.

Only thing I do not like the system is single round for Dramatic Scenes. I have ignored this, as in many Dramatic Scenes it just does not make sense. Thus I use multiple rounds for Dramatic Scenes and Action Scenes. In Dramatic Scenes each round contains simple subpart of the whole scene. F. ex. ball is split into several rounds as ball has time before eating, eating, chatging between meal, dances, and other events giving timing. I might even put events happening at end of each round representing something predetermined event like certain NPC giving out some information, or some NPC getting hurt, and so forth.

It takes a while to adapt to the totally different orientation of the system. And as DM (and other players) can suggest, some actions needs more than 1 raise to complete. Current Action Scene has one action which requries usually more than raise and that is causing Dramatic Wound. You can expand same to any other action you think more complex, but the point is you should allow partial building the result like in Wounding, if you do so.

The system is really bad for traditional tactical combat of Dungeons & Dragons. THe system is really bad for tactical roles of D&D-like gaming. On the other hand, system is totally superb for re-enacting and telling movie-like stories 7th Sea should create. This was the biggest problem in 1st Edition of 7th Sea: The system made it harder to get the feeling of movie like Heroic Swashbuckling into the game.

Yours, Kautsu