Zoom in on a frame of each and you will also notice the "jaggies" are more pronounced with streamclip and Neo-HD when compared to Pro-res. The pro res clip looks cleaner overall, and sharper.
Interesting. I checked my test clip which has quite a bit of straight lines and didn't find the streamclip to have extra aliasing compared to the plug-in. I'll try view it on the 55" and when I get back home have it tossed on the production monitor.
Are you comparing the clips in FCP or using QT player? If you are viewing in FCP, you also have to change the field dominance to "none" for the streamclip clip, as it will add aliasing in a 1080p sequence with the upper dominance default MPEG Streamclip uses. Just a flaw that MPEG Streamclip has.
Just for kicks I did a quick search online on various methods and not surprisingly found that views weren't universal but one stood out:
http://xrl.us/bg2gg3
Seems this person thinks the Reds are bleeding with Compressor. Which is plausible seeing that Apple reads color space at 0-235. I guess this needs to be tested more but I won't bother 
Also another thing to consider is the archived files. If the archives don't have the .thm and other data from the camera, the plug-in won't work without messing to trick it. So it's not a feasible solution for the first 1.3 years of clips captured WITHOUT this plug-in. I thought they'd have figured a way to do this. So capturing .img/iso of the CF seems to be the best way to archive.
Im not sure why people are so caught up in the speed game anyway. We spend all this money for the best quality of video, yet we compromise quality for speed in editing or transferring? Doesent make sense to me.
There's obviously not just one way to do this and many people will have preferences based on their needs. A buddy tested ProRes vs DNXHD and he liked the Avid better. Another guy loves using Compressor to batch process into Pro Res over the plug-in, which once set to use 6 or more cores (depending on the speed of the cluster connection) is as fast or faster as the plug-in but no timecode again.
Quality aside, if the speed difference is 10-20%, I'd keep it in FCP for the proper time (actually it uses the metadata from the camera's own clock for timecode). However, with the limitations of the file directory requirements, the super slowdown speed (up to 100% slower than streamclip at times) it becomes an issue when you have multiple clips. If I have 2 hrs of rushes, that'd take up to 4 hours to convert with FCP and 2 or less with streamclip. So yes, it can make a huge difference in batch. Also the streamclip files are smaller by a few percent.
On the 5D2, a fast card may not make as much a difference as on the 1D4, where it can squeeze an extra frame or 2 in burst mode and clear the buffer faster. Now is that worth the price premium? That's up to the user. Download speed is important for those on location. Clearing the cards for the camera in multi-card shoots means fewer cards needed and less running around. Express Card readers are excellent for UDMA cards for that purpose. Just don't buy that Delkin one as it SUCKS!