I think we broadly agree then. Once laws and customs are implemented properly all fortresses could have dwarves with different legal castes, classes and so on. This feature doesn't exist yet, l but it likely will. Customs will likely be procedurally generated which could lead to significant differences in gameplay between different civs, even for the same starting scenario. Starting scenarios would then allow you to focus on building a site for a specific sub-set of the population (prisoners, miners, people fleeing religious persecution, whatever), which alters the options you have available at the start, but keep site evolution open.
In my view, starting scenarios should resolve questions like:
What is the legal relationship between your site and the parent civ. What particular obligations does each party have? What happens if these obligations are not met? At some point you should be able to renegotiate whatever arrangements were originally in place.
What is the diplomatic status of your site? Are you on good terms with your host civilization? Will they send caravans? Military aid? Tribute? A siege?
Who will migrate to your site? Why? Are they subject to any restrictions? Are they permanent or temporary migrants? What do you need to do to attract a different migrant mix?
Are you bound to the laws and customs of the parent civ? What happens if you attempt to break with tradition (legalise cannibalisation, for instance)? Can you declare independence? What will the host civ do about it?
Who is administering the site? Is it a noble, pioneer, fleeing convict, cult leader, wizard, democratically elected mayor? Will the parent civ attempt to intervene in site development to steer things in their preferred direction? What will they do if they don't get their way?
Embark scenarios could define different initial states for all these question, while the player determines how things go from there. The same underlying mechanics apply to all sites.