Hi Steve,
From my experience building a remote site obs, the most important thing is having/learning a wide array of skills (from concrete/electrical/building/thermal stuff to telescope alignment/camera adjustment to software/network config). In some ways this is the most daunting aspect, you need to know enough about a lot to make it all work, but I suppose this is what makes it so rewarding in the end.
I guarantee there will be something you don't know enough about to fulfill the mission, that is when forums like these become invaluable, but better yet is having a partner in the whole affair, ideally another motivated astroimager with complementary skills that you can get along with.
Be prepared for sticker shock...no part of this is cheap and if you skimp now, you will pay later. The non-astronomy equipment costs are the real surprise, the comms/solar/site prep/building/tools/insurance!/transport!!/accommodation during build phase etc, can add up to more than a modest astro rig. Look seriously about scope hosting to make sure you are going down the right path for you.
As to your specific points:
1. 3G is far more reliable than satellite and costs a lot less. If you can get 3G that is good. Satellite can work (it's all we had in Ark up until recently) but you cannot operate the equipment directly, something like ACP needs to drive everything including starting up/shutting down. The busiest/laggiest time on the satellite is at dusk, and you are not reliably awake/coherent at dawn. Whenever you are in a rush the connection will drop or the latency will go above 2 seconds...it can be frustrating. You can transfer large files overnight with good bandwidth for relatively cheap (see SkyMesh Owl plans), but that needs some kind of automated/scheduled FTP synching software. (I wrote my own...) Plan where the dish/antennae is going to sit with a good signal, it will need a post and footer, and a way for the cable to go back to your building/obs.
2.
http://www.digital-loggers.com/lpc.html This thing can be driven by ACP via scripts, accessed over the web via satellite (not all devices can handle long latency connections). One thing to watch out for, the plugs on it are North American, lucky me being Canadian I had a bunch of suitable IEC plugs handy.
3. Size the system generously, you need enough power to get your always on equipment through 3? maybe more, overcast days. You normally don't want to be cycling the batteries down more than 40-50% per day or they won't last 3 years (N.B. even with with moderate cycling 10-20% they only go ~7-9 years). The batteries will cost a fortune, we have 20 x 105AH, I would have gotten more if I could have. Getting this part of the project is crucial since oversizing is expensive, and undersizing is disastrous and expensive later. I made a solar model spreadsheet with expected power draw (constant and nightly), sunny days, sun hours per day, sun angle over the seasons, battery capacity and solar panel capacity etc...to figure it all out.
The inverter needs to be very oversize, the Scopedome (and probably other) motors draw a LOT of current to start up. We started with a 'high end' portable Projecta 12V 600W inverter (or 1000W? can't remember, we were going to use one each), it would stall opening the shutter which is only a 120W motor, with nothing else on. We now use a 48V Latronics 1800W inverter (5400W surge), it doesn't stall...
Get a MPPT type charge controller (we use Morningstar-Tristar60) it easily pays for itself many times over in less required panel capacity and battery life which are the real cost drivers along with the infrastucture around them.
N.B. 48V system is far superior to 12V, for the charge controller, the cabling, the fuses, the inverter, the battery setup etc.
Plan where to put all these batteries, fuses/electronics, not trivial actually, and how to mount the solar panels, which need more footers and posts with special brackets.
4. GPS is nice but no big deal. I have a Garmin 18x LVC setup with NTPd on windows, works fine, under msec precision as long as you can get a compatible USB-RS232 converter, this was trial and error.
5. We are using AAG, it works as advertized, software setup is good, their service has been good. For example the rain sensor failed after 4? years, they sent a replacement for free, we just haven't gotten around to installing it yet, but we don't worry about sudden showers in Arkaroola, so your situation is perhaps different. The Boltwood is surely better but cost about 5x more. Either type need a post to mount on and cables to run back to obs (next to satellite dish is our solution).
6. No question, FIT-PC. Between Mark and I we must have ~7 FIT-PC2's (lost track now). Mark is trying the new fitlet, looks really good. Fitted with an SSD there are VERY low power, impervious to temperature, industrial grade, wake-on-lan (fitlet), auto-boot on power cycle etc, ideal for remote use.
7. Nice to have inside the dome, although I find the audio more useful, you can tell by the sounds if everything is working right...Getting stuff to work across firewalls is a real pain (some cameras use activex plugins), and the bandwidth video consumes is startling, use sparingly.
8. Don't have moisture in Ark, so can't help there. I'd love to have aircon to bring the equipment temp down in the dome before sunset, but I would need a truckload of batteries...so opening the shutter early is the cheaper solution.
9-10. Land owner being around is your best defense.
11. One of those hidden costs not necessarily budgeted for, good thing it is on your list.
One other suggestion I would make: If at all possible set everything up that you can in your backyard and debug, for many weeks. Ideally this would be the dome,scope,computers,routers,camera s,software that you are going to use (pier/footing/solar excluded, they are to hard to move).
Best,
EB