I usually keep the desktop booted into Windows, and the laptop booted into Linux. Steam's network streaming feature comes in handy, with that setup. There's no reboot involved, but I *do* end up using two computers at once. Good thing the wife has her own laptop...
My point wasn't that your demographic doesn't exist, just that it's nowhere near low-hanging fruit. Being generous, it's the intersection of gamers and people who use Linux in any form. Realistically, the demographic is more nuanced, something like "Linux users who have an x% likelihood to buy a specific Windows-only game, and an increase of y% that they'll buy it as a Linux port". There's a point where the splinters of markets get too small for a lot of developers to worry about, where it becomes a question of spending 5% more in development to gain an extra 2% in sales, and it doesn't make sense any more.
I'd expect Humble Indie Bundle numbers to be fairly high, in terms of Linux user representation. I know that I've marked "Linux" as my OS of choice every time I've purchased one. Out of 4,998,506 purchases, 332,944 have been Linux (6.7%), 616,596 are OSX (12.3%), and 3,962,890 have been Windows (79.2%). Linux users generally pay nearly double what Windows users do, so there's that in their favor.
~20% of the market may be worth it, depending on how much more work it takes to get there. ~5% may not be, based on the same factors. Most of the large developers seem to be taking the gamble that enough of the people that are "Linux users" will still buy a Windows-only game (or buy the same thing on a console). And while there are exceptions, they seem to be mostly right. From the dev's perspective, losing a few percent of your market is bad; spending a disproportionate amount of your budget to pick up a few extra percent is worse.