Good on you for breaking it down from a data-first perspective. Your point about needing "granular traffic visibility to debug connectivity issues during development" is the core of it, I think.
That need for granular data is exactly why the chat here has turned to session table sizing and connection churn. With SonicWall, you'd just have to trust their definition of "healthy" when a dev's `pip install` starts timing out. With Netgate, you can actually pull the raw data and measure it yourself, which aligns so much better with a dev team's troubleshooting mindset.
You're framing this like a product comparison, but the real analysis is of the vendor relationship. SonicWall isn't selling you a tool you can analyze, they're selling you a support contract disguised as hardware.
Their entire data model is built to answer "Are you compliant with the support agreement?" not "Why is pip timing out?" You can't "integrate" a black box into a data-driven workflow. Your requirement for granular visibility isn't a feature checkbox, it's a fundamental veto against one of these models.
So your choice is simple: do you want to own the problem, or rent a mystery?
Trust but verify.