Well, in order to integrate DLC properly with a game you do have to do some planning ahead of time so that the code supports the hooks. Remember how the official mods were handled in Oblivion, with the character receiving a bunch of notes from invisible messengers as soon as they were in the game? With an understanding that you are hoping to release DLC you can build the engine with DLC in mind, and provide semi-visible adventure hooks such as what's being discussed for DA:O.

From a technical standpoint, the mainstream audience wants to see engine changes from release to release, not just a new story. So ME2 isn't just the story sequel to ME, but a whole new engine, or at least a heavily modified version of the ME engine. From a developer standpoint this is a huge waste. Unless you build a standardised engine that you can license out as Valve did with the Source engine you're wasting a lot of man hours on minor tweaks that will be forgotten as soon as the next game comes out. DLC allows you to get more bang for your engine "buck" as well as keep players interested in the franchise while you do work on the next game.

Besides, if you were running the company, what would you do with all of the artists, designers, modellers, et cetera while the main game was being tested and finalised? Send them home to sit on their thumbs? Have them play the game 40 hours a week "testing"? Fire them all and hope that they haven't found a new job when you need them again? Or set them to working on an optional side-project which could turn into added revenue in the future?

If gamers were happy with the way games looked 20 years ago then games would still have production times of 6 months. In the modern world that doesn't work, and games have longer development cycles than blockbuster films. You can't expect companies to ignore business opportunities like planned DLC in an environment where they have to invest tens of millions of dollars into a project years before it will be released.