"Imagine a measure quantifying for instance the length L of a body. Is then a difference between (L) and [L]?"
I think you actually mean to say [a,b] and (a,b), referring to closed and open intervals. There's no length without interval. Your sets (L) and [L] are dimensionless.*
"Why cannot point-set topology perform a cut A|B between IR and IR- without an arbitrary choice between -)[, -](, -), 0, (, or undecided? Neither of these variants deserves to be called a symmetrical cut."
With due respect, Eckard, I don't think you understand Dedekind cuts. The interval of successive integers is most certainly symmetric. I tried to explain this by the example that in Dedekind cuts, there exist two definite numbers that when multiplied together equal the square root of 2. This is by way of proving that "the least of the most and the most of the least" in the continuum have boundary of length 1 and measure of zero -- since we cannot say what the two numbers are.
* I touched on this in my ICCS 2006 paper. First paragraph of 7.0, the discussion section: "The positivity requirement (5.11.1) and the infinite variety of state spaces made available by the Hilbert space -- along with the constructed continuity that unites real and complex analysis in a backward-forward projection between S^1 and S^3 suggests the preservation of equilibrium on the intervals (0,1) and [0,1] in a true transformation of an indefinite and continuous measure space to a discrete counting function symmetric about the complex plane axis." (Ray, 2002)