If you want added redundancy, the 2nd power supply and/or Raid Hard Drives are a plus.
Most hosts deploying a large amount of servers prefer to put money into either redundant hardware or spare servers rather than spend extra money on upgraded warranty plans which would still result in longer downtime.
A Y-Cord runs both cables from each power supply into a single connector. This allows you to use remote reboot on both power supplies, but doesn't offer actual redundancy at the power strips.
The Dual Cords let you run each cable to a different power strip or to a different UPS (if you're in a facility without inline UPS) so you have full redundancy. However you can lose your ability to do remote reboots when doing this (however you could buy a DRAC4 management card and reboot your Dell with that).
Onboard is fine, occasionally I have bought Intel Dual NIC network cards for specific servers needing additional ports or to offload some of the load. I seriously doubt you'll see any negative performance with them onboard, even if you did it's nothing you can't add later during a maintenance period to evaluate.
A firewall at the software level will just slow down your server, and likely cause a lot of extra work per machine. You can probably do without for a little while as you start out, and then start looking at hardware firewalls, plenty of good recomendations on the forums here if you do a search.
Alienware's server cost 2x as much as the same hardware from Dell. Do you really think they're using different parts or built that much better? Their Hivemind 1220 is a Supermicro design, I'm not sure about the other 1U/2U as they're covered with the custom Alienware logo.
You're better off getting a Dell and slapping an Alienware logo on it.
Also to the comment "Most hosts deploying a large amount of servers prefer to put money into either redundant hardware or spare servers rather than spend extra money on upgraded warranty plans which would still result in longer downtime."
Are you saying its better to buy more spare parts than to upgrade the expensive warranty plans?
If you spend say another $1k-2k on a spare server, you have a fully ready server to deploy, could just pull and install hot swap hard drives into it then swap cables and you're good to go in less than 10-30 minutes. Then call in for a RMA with your 3yr basic next day warranty on the server that's down.
The alternative is paying another $200-300 just to get same day coverage either 8hr, or 24hr. Not a whole lot extra, but it adds up, and is pretty redundant if you already have on-site coverage and parts. Plus scheduling and arranging for a dell tech to repair or deliver a replacement can be a problem.
Their silver and gold support get into the $1k-1.6k range. Clients have had this before and I've called in on it, nice to get past the queue extremely quick but often you're just confirming with them, that yes in fact a hard drive is dead and you need a replacement... Would have rather saved $1k.
In 3yrs, you're really only likely to suffer one critical failure if any, usually a hard drive. Sometimes memory or a power supply.
For critical servers, I do upgrade to the 8hr or 24hr coverage. For everything else I stick with basic and pickup a few spare parts to keep on hand. It's also nice as you'll have them on hand for upgrades if a customer wants extra memory, larger hard drive, extra hard drive, etc.
Just think over your situation at your colo, the access you have, the level of downtime you can afford, how you could even arrange for a dell tech or delivery to be made, the risk, the costs, and see how the pro's and con's work out for your particular situation.
On redundant power supplies, think about how your servers will be accessed. In the web serving area common practice is to forego redundant power supplies and most other redundancies at the machine level since (in a large environment) there is redundancy at the network level (load balancing).
If your game servers do not offer failover, then it makes sense to add redundancy at the server level. If you run multiple game servers for the same game and there is redudancy in the software/protocol, then you will want to save money by not adding redudancy at the server level.
Keep in mind though that network outages are much more common than power outages (either problems with the colo power supply or a power supply dying), and especially for gaming it might make more sense to spend that money on additional redudancy at the network level (either in your cabinet or at the colo) rather than on power.