I put hill/mountain, but it clearly depends on your goals. Hill/mountain is the easiest to defend, and has good early production. Windmills are basically useless in the grand scheme of things. If you have the hammers to build windmills, you need them for public schools, or unit, or archaeologists.
Think about it: Late-game buildings that add incremental value never pay off. The only good thing about a windmill is the specialist slot.
250 hammers, +2hammers, +10% when building buildings.
Ok, let's say, for the sake of argument, that you're at 50 production when you get economics. +2 hammers is 4%. So we'll call it 15%. By the time you can build Windmills you should be building Workshops, Factories, Public Schools instead, but let's see how it affects things to build a windmill first: A public school is 300 hammers. 6 turns at 50 hammers. Building a windmill first delays that by 5 turns, but adds +14% hammers. 300/57.5 = 5.22 turns. So 6. So, you've delayed public schools by 5 turns. But maybe it evens out after factories? You've saved 1.66 turns... realistically just 1. And after banks and research labs? 3.4 turns total... After stock exchanges you've *almost* broken even. And the game is over. Grats you!
Plus, if you're going for growth, plant your city on the one tile that can't grow food!
So, hills > flatland, period. The only question remaining is river vs mountain? Well, this depends on whether you want a water mill (optional IMHO) and a hydro dam. (nice to have but comes late and *not* essential*)
For me it's more about the surrounding tiles. Which location gives me the best overall tiles to work sooner? (and later)
Other tiles all being equal, I'll take that river/hill over mountain/hill if I don't think I need to defend the city, and I don't think it'll grow into a huge population. (Observatories are AWESOME in *high pop cities*)
But, flatland? Only if it nets me something awesome like a wonder, or I can settle on a lux to immediately solve a happiness problem I would otherwise have.