This is a lovely idea in theory, and I wholeheartedly support it.

In the real world, however,

1) ordinary users (i.e. the people in other companies who might want to use our company FTP site) are probably not going to have any programs installed that can talk to SFTP/SCP etc., and are not going to want to bother their IT department with it 'cos it'll probably take about 6 months to get any software approved and the project will be a dim and distant memory by then.

2) these same users, even if they had the software, are going to be blocked from the vast majority of other methods of sharing information with us (including conventional FTP on a different port) by overly-draconian internet filtering policies, probably mandated by the same people in the IT dept. that insist that all their FTP/SFTP/etc. sites are locked down and secured with all types of encryption and on a strange port.

Sadly, there are few practical alternatives to lowest-common-denominator FTP access for communicating with a lot of companies.

That is true, but even if it is the case that the date that counts is the US publication date and the foreign publication date simply has to be disclosed, my (limited) understanding of the argument in the complaint is that they didn't disclose the foreign publication date at all. This would mean that they would be alleging that the copyright registration was fraudulent in any event.

That said, from my reading of the complaint, the arguments substantially depend on the idea that it is this initial (foreign) publication date that counts. This leads to two possibilities: a) the lawyers representing Mr. Shirokov are idiots, or b) it's the April date that counts. I am prepared to give Mssrs Booth and Sweet of Booth and Sweet LLP the benefit of the doubt in the absence of any evidence to the contrary...

Of course, as mentioned in my previous post, IANAL, and I would welcome any contrary evidence :).

From the complaint:

45. An application for a registration of copyright in a published work requires a statement of the date of first publication; the nation of first publication should also be given. Specifically, under the Copyright Office’s guidelines an application covering a work first published outside the United States should state the date of first publication there, and should be accompanied by a copy or phonorecord of the foreign edition as first published.

IANAL, but whoever wrote the motion is, and they seem to think that it's the date of first publication of any (presumably Berne) treaty country that counts.

I don't know if it will help you, as you and your user's needs may be quite different to mine, but this is the set-up I use for a small company that I do part-time IT stuff for. It works pretty well for them so it might work pretty well for your lot too:
This is for a small co. of about 12-15 engineers, (depending on how many part-timers your count). They do lots of computational modelling, so lots of storage space and CPU is needed. This sounds like it might be a bit similar to your needs (if you're going to do anything interesting with that large collection of videos & multimedia)

They have one "main" file server for project work. It's a white box PC from the local shop and has a couple of TB of hard drives in it. It runs ubuntu server with samba (like all the other servers in this co.). It has needed work about twice in the last 4 years.
They have a couple of old Dell workstations which are too slow to do engineering on now. One runs an external-facing FTP server (it could probably run a small website if needed, too) and one runs an internal wiki and a few other similar tools. I could probably move some of the internal stuff into the main file server, but we had the old machines kicking around, and it's useful to be able to fix stuff without breaking the whole network for the whole company at the same time.
We have a modelling file server, which is a big Supermicro rack server. It's a 4 or 5U box, because they have an open-plan office and nowhere to put the rack, so the rack-mount servers have to be very big (for what they are), so they can be quiet. This has space for 8 hard drives so you can pack it out with largish drives and there will be enough space for all but the most data-hungry organisations. It's expensive compared to the white-box PC, but if you really need the extra space, it can be difficult to find an off-the-shelf machine with space for more than 6 hard drives (and it's a lot easier to replace one if one fails, too).
We have a backup file server. This uses rsync to mirror the (newer) contents of the other two servers, so that if one of them falls over, we don't have a bunch of engineers sitting around while I get the train into the office, work out what's wrong, get the right part, fix it, etc. It also compresses the important (non-replacable) data every week so that someone can copy it to an external drive and take it off-site. Much cheaper than the internet connection that we would need to mirror a week's work in reasonable time over the internet.
All of the computers are cheap white-boxes from the local shop running windows XP or 7 with various versions of MS Office (whatever was current when the machine was bought). No-one seems to have any problem with the fact that the boss uses XP and office 97 while the new guy uses Win 7 and office 2010, and I have better things to do than make an issue out of it. We keep track of whose license is who's on the wiki. Most machines also have OpenOffice, but there is general user resistance to that concept.
We have a couple of PCs for doing number crunching. They sit in the corner and run VNC servers. If people need to crunch numbers they use them, otherwise they use their own cheap workstations.

In summary: buy off-the-shelf PCs for the users. So long as they have windows, office, anti-virus, etc. they'll get on with what they need by themselves. The hassle of getting people to use linux or OpenOffice is not worth the 250 pounds we pay to MS per computer. An off-the-shelf workstation or server with some extra HDs and some version of linux makes a perfectly adequate file server. Use sneakernet for backups.

As I say, your situation may be completely different, but I hope mine might give you some ideas.

While OpenFOAM is certainly really powerful, it is short of a GUI* (except for results-visualisation), and might therefore be less than ideal. That said, it is simple enough to use with a walkthrough, and the fact that the interface is basically composed of text files should make it easier for students to get back on track if they go wrong (this is a big problem, for example, with teaching CFX). If the main focus of the work is going to be running an essentially pre-built model (which the students then rebuild and run) and looking at the results then it will be more than capable. It is also easy to investigate the effects of changing discretisation methods, solution schemes and the like. If you want it to be easy for students to perform tasks such as changing the angle of the aerofoil relative to the mesh, change the mesh resolution, etc. it may be less ideal as this will involve lots of messing around on the command line (and in text editors).

* In fact, there is a GUI available from symscape.com, but this costs money (albeit not a lot). (no affiliation, I've never used this software and can't speak for its quality)

Now, I'm not saying that anything in your post is either wrong or right, but I feel I have to point out that if your best reference for a science story is the Daily Mail, then this is an argument that you are not going to win.

What other products come with a legal document stating that, if you so choose, you can return a single component for a refund?

True as this undoubtedly is, I think this is the wrong attitude to take. Simply saying, "OK, Mr. Government, if you want to block bits of the internet go ahead, we'll just work round you." gives the impression that they have the right and justification to censor bits of the internet at will and it's up to us to work round that.

While the sort of people who read slashdot are able to circumvent this kind of thing, does that make it right to censor the internet for the rest of the less technically savvy population?

This is the first (and I hope the last) time I have heard such a scenario described as "insidious".