Well I'm intrigued now. I haven't found an authoritative source yet, but I have founds lots of confidently expressed positions that, reading between a few lines, suggest that's a big oversimplification. Mostly extrapolating from the graphic here but consistent with some narrative elsewhere found in 10 min of googling. Allow me to back blindly over the dead cat a few more times with some inferences from what I'm seeing:

- It just makes sense that there is a difference between unconstrained rotation and slightly constrained. i.e. the cleat doesn't release in a nano-degree, but rather moves through some level of constrained rotation in the process of disengaging. Call it cleat taint -- taint float, taint release.

- It seems Crank Bros are very consistent in saying that their float (i.e. unconstrained) is 6 while their release angle (i.e. float + the extra constrained rotation) is 15 to 20.

- From the pic it looks like their 6 deg of float may be of total angle, i.e. +/- 3 deg to either side

- Extrapolating a bit, it looks like the 15 or 20 deg release angle may be one-sided. i.e. if you consider the awkward heel in release, you'd have a 30 to 40 degree total angle between heel in and heel out release.

- if that's right, there's on the order of 12 to 15 deg of taint. You're retained but constrained and not floating. My heel flick is not calibrated but staring at a protractor that doesn't look unreasonable.

I haven't found much distinction between float and release angle elsewhere. And this is one of the few cul de sacs of bike nuance that Sheldon did not dive deeply into before his time was cut short, RIP. If any Tech God knows, perhaps you can illuminate us.