Help with Workstation build!

Hi there,
I'm new to this site but I think you guys are great. I want to build a workstation that is cpu heavy and can run Maya autodesk, Arnold renderer (cpu renderer), and some Zbrush and mudbox. Basically I want to cut my render times down. I have a budget of around $5,000 and I came up with this list from Pc Part picker.

2x Intel Xeon E5-2660 V4 2.0GHz 14-Core Processor
CPU Cooler
2x Noctua NH-D15 82.5 CFM CPU Cooler $88.49
Asus Z10PE-D16 WS SSI EEB Dual-CPU LGA2011-3 Motherboard
2x Kingston ValueRAM 32GB (1 x 32GB) Registered DDR4-2133 Memory
$154.99x2= $ 309.98 (maybe 128gb, 64gb each processor)
PNY CS1311 960GB 2.5" Solid State Drive
Western Digital BLACK SERIES 2TB 3.5" 7200RPM Internal Hard Drive
Video Card
NVIDIA Quadro M4000 8GB Video Card
Phanteks Enthoo Luxe ATX Full Tower Case
Power Supply
EVGA SuperNOVA 1000G2 1000W 80+ Gold Certified Fully-Modular ATX Power Supply
Optical Drive
Pioneer BDR-209DBK Blu-Ray/DVD/CD Writer
Operating System
Microsoft Windows 10 Pro OEM 64-bit
Extra Fans
4x Corsair Air Series AF140 Quiet Edition 67.8 CFM 140mm Fan
Corsair Air Series SP120 Quiet Edition (2-Pack) 37.9 CFM 120mm Fans
The total is around $5500
Ok these are the question that I have.
The motherboard Asus Z10PE-D16 WS is around $450 and the Asus Z10PE-D8 WS is about $550. What would be the difference between these two? Do they both support m.2?
Is it possible to run a single e5 2660 v4 and than just another e5 2669 v4 later?
Also with the memory, it seems like a lot is better but by Intel e5 2660 v4 says that the max memory bandwidth is 76gb/s
which would make 64gb the most it could take. Right? Maybe?
I'm trying to go for the most memory for the least amount of money. Is there a difference between a ecc ddr 4 32gb memory that cost $400 than $150?
What's the difference between ECC registered memory and ECC unbuffered memory?
Is it worth it to expand the memory from 32gb to 64gb per processor?
The Quadro video cards are good for Maya autodesk but Maya hasn't certified the Quadro m4000 yet. I think that because it just came out. The Quadro m4000 is the new version of k4000. Is that correct?
Last question, the case to fit the motherboard. Pc part picker suggests a Full tower ATX and the motherboard is a SSI EEB. Is going to work? It seems I have more options with the ATX style case than the SSI EEB.
Thanks Everybody for time.

OK, some things to consider: You want to have 4 memory dimms per CPU to take advantage of quad channel. You are probably better off with 120mm coolers since space between CPUs is limited. SSI EEB basically fits into most E-ATX cases. I know my Nanoxia Deep Silence 5 Rev. B could handle that.

Also I guess money isn't critical here and the focus is just maximum performance. But I'll say this: going with dual E5 2670 (v1 that is, Sandy Bridge with 8 cores 16 threads each running up to 3GHz on all cores) will cut your expenses for CPUs by 90%. (Yes. Ninety.) Those chips are on ebay for 50-70 bucks. Here is what I plan to do with those: CLICK

On to your questions:

The differences of those boards should be pretty obvious. Here are the specs of the D16 and here is the D8.
The D16 has IP management, more memory, less PCIe. The D8 has less memory, more PCIe, no IPMI.

Single CPU: Wendell did a video on the D8 and as far as I remember he used only one chip. But I am not 100% sure on that.

Memory: Amount and bandwidth has nothing to do with each other. The D16 has written in it's specs that you can put up to a terrabyte of DDR4 onto that thing. Bandwidth is basically 'how much stuff can go through the memory per second'.

Registered memory is available in higher capacities and electrically slightly different from unbuffered dimms. Also registered memory is slightly slower.

I leave it at that for now.