"requiring additional steps than a native app making equivalent OpenGL calls would demand"?
I see what you're getting at here, but I don't know if that wording works in this context. The problem is that all WebGL implementations require additional steps compared to native. A simple example: WebGL implementations validate index ranges before attempting to draw, while native apps don't. That's a small difference that shouldn't impose too much overhead, and it obviously falls outside the type of examples that you brought up, but it does match the wording proposed.