Reduced timestep of high resolution simulations from 50 to 40 fs because of instability in some systems

Fixed issue affecting systems with long polymers where update_splines previously used float16 to obtain unique beads; now rounding to nearest 1/10th nt

Improved orientation potentials at junctions when local_twist is enabled; now dihedrals are more local and work in more situations

Made it so a bead can be located at multiple distant locations within a segment (e.g. for looped constructs)

1) Fixed lattice_type setting for cadnano structures, affecting ENM
2) Raise exception for circular segments (sites that loop such as RNA stem loops remain problematic)
3) Made translate and rotate functions return if nothing is selected by filter

Fixed corner case bug where crossovers at end of ssDNA were not neccesarily recognized as intrahelical connections

now if the bead on the central helix is the same at each crossover, an angle potential is applied instead of skipping the crossover

Slightly reorganized simulation protocol; added options to mrdna script and simulate.py for using twist in the coarsest step

First, ssDNA segments are now broken at crossovers
Second, ssDNA segments aren't allowed to stack intrahelically