Recommend a Modeling Approach for a Multi-Source Enterprise Warehouse

You're advising a large insurance company building a new central data warehouse. Context: they have 40+ source systems (claims, policy admin, agent commissions, several recently acquired subsidiaries each with their own legacy databases), regulatory requirements to demonstrate full audit trails of how any reported number was derived and from which source, a small central data engineering team, and business analysts who need to self-serve reporting on claims and policy metrics without waiting on data engineering for every request.

1. Would you recommend a pure Kimball star schema, a pure Data Vault, or a layered combination? Justify against the specific constraints given (40+ sources, audit requirements, small central team, self-serve analyst need).
2. Sketch the high-level layered architecture (name the layers and what lives in each) if you recommend a combination.
3. A subsidiary is acquired eight months into the project and needs to be onboarded with a different, incompatible legacy schema for "policy." Walk through what changes at each layer of your proposed architecture, and explain why this scenario is specifically what makes a pure star-schema-only approach risky here.