nbody_by_type.py reads a LAMMPS data file (or an excerpt of a LAMMPS)
    data file containing bonded many-body interactions by atom type
    (and bond type), and generates a list of additional interactions
    in LAMMPS format consistent with those type (to the standard out).

    Typical Usage:

    nbody_by_type.py X < old.data > new.data

    --or--

    nbody_by_type.py X \
                     -atoms atoms.data \
                     -bonds bonds.data \
                     -nbody X.data \
                     -nbodybytype X_by_type.data
                     > new_X.data

    In both cases "X" denotes the interaction type, which 
    is either "Angles", "Dihedrals", or "Impropers".
    (Support for other interaction types can be added by the user. See below.)

    -------- Example 1 -------

    nbody_by_type.py X < old.data > new.data

    In this example, nbody_by_type.py reads a LAMMPS data file 
    "orig.data", and extracts the relevant section ("Angles", 
    "Dihedrals", or "Impropers").  It also looks a section named "X By Type",
       (eg. "Angles By type", "Impropers By type", "Impropers By type")
    which contains a list of criteria for automatically defining additional 
    interactions of that type.  For example, this file might contain:

    Angle By Type

    7 1 2 1 * *
    8 2 2 * * *
    9 3 4 3 * *

    The first column is an interaction type ID.
    The next 3 columns are atom type identifiers.
    The final 2 columns are bond type identifiers.
    The * is a wildcard symbol indicating there is no preference for bond types
    in this example.  (Optionally, regular expressions can also be used to
    define a type match, by enclosing the atom or bond type in / slashes.)

        The first line tells us to that there should be a 3-body "Angle" 
    interaction of type "7" whenever an atom of type 1 is bonded to an atom
    of type "2", which is bonded to another atom of type "1" again.
    The second line tells us that an angle is defined whenever three atoms 
    are bonded together and the first two are of type "2".
    (Redundant angle interactions are filtered.)

        New interactions are created for every group of bonded 
    atoms which match these criteria if they are bonded together 
    in the relevant way for that interaction type (as determined by
    nbody_X.py), and printed to the standard output.  For example, 
    suppose you are automatically generating 3-body "Angle" interactions using:

    nbody_by_type Angles < old.data > new.data

    The file "new.data" will be identical to "old.data", however the
    "Angles By Type" section will be deleted, and the following lines of
    text will be added to the "Angles" section:

    394 7 5983 5894 5895
    395 7 5984 5895 5896
    396 7 5985 5896 5897
     :  :   :    :    :
    847 9 14827 14848 14849

    The numbers in the first column are counters which assign a ID to 
    every interaction of that type, and start where the original "Angles"
    data left off (New angle ID numbers do not overlap with old ID numbers).
    The text in the second column ("7", "9", ...) matches the text from the 
    first column of the "Angle By Type" section of the input file.

    -------- Example 2 -------

    nbody_by_type.py X \
                     -atoms atoms.data \
                     -bonds bonds.data \
                     -nbody X.data \
                     -nbodybytype X_by_type.data \
                     -prefix "SOMESTRING" -suffix "ANOTHERSTRING" \
                     > new_X.data

    In particular, for Angle interactions:

    nbody_by_type.py Angles \
                     -atoms atoms.data \
                     -bonds bonds.data \
                     -nbody angles.data \
                     -nbodybytype angles_by_type.data \
                     > new_Angles.data

    When run this way, nbody_by_type.py behaves exactly the same way
    as in Example 1, however only the lines of text corresponding to
    the new generated interactions are printed, (not the entire data file).
    Also note, that when run this way, nbody_by_type.py does not read the
    LAMMPS data from the standard input.  Instead, it reads each section of
    the data file from a different file indicated by the arguments following
    the "-atoms", "-bonds", "-nbody", and "-nbodybytype" flags.

    "Angles" is a 3-body interaction style.  So when run this way, 
    nbody_by_type.py will create a 5 (=3+2) column file (new_Angles.data).

Note: the atom, bond and other IDs/types in need not be integers.

Note: This program must be distributed with several python modules, including:
        nbody_Angles.py, nbody_Dihedrals.py, and nbody_Impropers.py.  These
      contain bond definitions for angular, dihedral, and improper interactions.
      (In case any new interaction types are ever added to LAMMPS, with only
      a few lines of python it is easy to edit to define new bonded 
      interaction types by supplying new "nbody_X.py" python module.
      Refer to the modules listed above for examples.)

Note: Optional "-prefix" and "-suffix" arguments can be included to decorate
      the interaction IDs (first column).  For example, -prefix "auto_" and
      -suffix "_angle", causes "new_Angles.data" to contain the following text:

    auto_394_angle 7 5983 5894 5895
    auto_395_angle 7 5984 5895 5896
    auto_396_angle 7 5985 5896 5897
          :        :   :    :    :
    auto_847_angle 9 14827 14848 14849