A few thoughts... CR did a review on testing lenses a year or so back... it concluded that one lens could have different levels of AFMA needed on multiple bodies... It also said each camera could be fractions of percentages off and still be "within specs" and lenses the same way... so what may be perfect with one lens/camera combo may or may not be perfect on an other camera or identical lens. AF AFMA is definitely worth your time to check out... it may be right on, it may not, but it's something to do before jumping to conclusions. Also, as far as soft focus, if you can use live view, MF the image to tack sharp images, then it simply means the camera is capable of a sharp image, however the AF is slightly off, so fine tuning it may be needed and in worst case situations, sending the lens/camera to canon so they can calibrate it and make it perfect for you. Lastly the level you have our AF tracking set up for erratic focusing may or may not have a factor... if the AF is constantly looking for new points to focus on, it it's constantly looking for changes, for "erratic" movements, it may be jumping on and off focus before you are capturing the image in the first place. Those menu's are relatively new so perhaps some experimentation with those may be needed to find the exact best setting for the environments you are shooting in.
I know there are reports in very very low lighting it may struggle because there is no focus assist light or any additional chip to power the drive, but i've also read reports the camera is razor sharp when configured properly. Even some canon ambassadors have said technically speaking, the metering and AF system is more sophisticated on the 5d3 compared to the 1d4 and the 5d3 SHOULD have better tracking and AF accuracy than the 1d4. In practice it may not be as clear cut but i'd recommend calibrating everything and testing everything in a controlled environment first before you throw it in the deep end at a concert when you're still learning the AF system.