There are places where a desktop is more suitable but that doesn't mean that within an enterprise they can replace a lot of them with other devices.
I agree but it's interesting that the 2 examples you picked were Warehouse and Salesmen. Niether of these are "office" enviroments in my opinion. A warehouse clearly isn't and, to be honest, most don't have desktops in them now. Maybe one or two to act as docking stations for the scanners but that's it. The desktop has never had a home in that enviroment so it's not being eroded in the way that people seeem to fear. Salesmen are out on the road rather than based in offices and portability therefore becomes the trump card. Of course, when that salesman decides to write a lengthy covering letter to one of his clients he'll probably sit at a desk to do it. When he just want to check his diary to see where his next appointment is he'll do it on his iPhone - in fact, he already is. so the desktop's home isn't so much the "business" as the "office". Until we see the office dissapear we won't see the desktop dissapear.

I think that's my point, it's not that the desktop world is the be all and end all of the computing world, it's not. But those areas that can naturally shift onto other platforms pretty much already have and the desktop is still here. Without another major shift in the hardware forms we're using I don't see a significant erosion of the desktop marketplace on the horizon.

if you're uncomfortable writing large documents with an onscreen keyboard (I know I would be) there is nothing stopping you from using a bluetooth keyboard connected to a tablet
That's true. I actually think there are two choke points for a tablet being used for general office use: screen size and input devices, both of which which could be resolved separately with "plug-ins". Of course, you're probably not going to want to carry a keyboard and pair of monitors around with you (at least in their current form) so you're then really talking about docking stations on desks. I can definetly see that happening but, if it does, we'll probably want a desktop orientated operating system to run on it.

If that's the case then MS's attempt to merge a tablet gui with a desktop gui would be right on the money because we're not just talking about an OS that needs to work across different devices, we're talking about an OS that needs to work differently on the same device according to the experience the user wants right now. From what I've heard of Metro (I'll freely admit that I've barely tried it beyond an occasional 5 minute shufle on freinds' machines) I don't think they've been particularly successful in their first attempt but this isn't the first time they've released a product where we all baulked at the UI (Office Ribbon anyone?) and they've got a track record of learning from their mistakes.

I'll write for what is here, at the moment, not for what I think will happen in the future
My philosophy entirely. There's no shame in being a year or two behind the curve.