prevent that LIGHTCONE_PARTICLES_GROUPS can fail because of non-unique IDs in the (replicated) particle set

added the LIGHTCONE_MULTIPLE_ORIGINS option, which allows one to select a different origin from (0|0|0) for each lightcone

fixed reading in of LptScalingfactor when SECOND_ORDER_LPT_ICS option is used combined with HDF5 input format

changed order of output field for accelerations such that old format 1 files can be read even if OUTPUT_ACCELERATION is on

fixed problem due to due MPI_IN_PLANCE in MPI_Alltoallv when MPI_HYPERCUBE_ALLGATHERV was enabled, added MPI_HYPERCUBE_ALLTOALL option, and more consistently wrapped MPI_Alltoallv, MPI_Alltoall, and MPI_Sendrecv into functions that control the communication algorithm and maximum message size

fix for hideous instability on some systems where MPI_Gather in a shared memory communicator can sometimes return permaturely before all data has safely arrived

fix problem that spurious additional snapshots can be created if code is interrupted right after snapshot dump

fixed a problem when the collective SUBFIND(-HBT) algorithm was working on particles on a lightcone (i.e. for LIGHTCONE_PARTICLES_GROUPS)

fix in yesterday's flawed change to when set_viscosity_coefficient() is called (thanks to Leonhard for pointing this out)

added a barrier so that the 'Processing overall took' log message reports the maximum time over all ranks