Resolve "Use a logscale for the QMC intrinsic scales"

Closes #75 (closed) The BSS intrinsic scales are now kept on a log scale which allows to go to much lower temperatures. For the few trial runs, I got the impression, that the stability is more or less temperature independent. It is however controlled by Nwrap and dtau, as Nwrapdtauphysical_scale set the stability between two stabilization points. Hence, even if Nwrap=1 one should be able to improve the stability by reducing dtau. I have carefully tested the updates as I changed core parts like cgr1.f90 and cgr2_2.f90 on all available and working examples (this excludes Hubbard_Mz_Square_Can and Z2_Slave as both seam to be broken in the current master branch). Also the stability of the example models did not change significantly. Before, in my model, I was limited e.g. to beta <= 4 for V=8. Now I can increase beta to 8; 16; 32; 64; 128 and 256 without any slightest sign of problems, i.e. no instabilities, no NaN, ...