merge PQEq to main branch

[KenichiNomura]

• the communication routine needs to be generic & extensible to handle different data type, number of elements etc for normal QEq, PQEq, spring force, MSD calc etc. This is done.
  -> new comm copy&move modes for the shell position. no needs for normalization and shift b/c they are relative from the core charge.
• the table construction for non-bonding interaction needs to handle the difference in QEq, PQEq, and the LG correction. Especially the PQEq cutoff distance is not always 10A, which affects to the linkedlist cell size and its construction in many places.
  -> the coulombic interaction uses tbl_clmb in QEq&ENbond, tbl_clmb_[ss,sc,cc] in PQEq&ENbond
  -> the LG correction should check element type
• the NEbond is quite different in QEq and PQEq. Have two separate functions (NEbond_QEq & NEbond_PQEq) and use procedure pointer depending on user input?
  -> use a logical flag (isPQEq) for now.

[KenichiNomura]

• How to read the PQEq parameter file?
  -> read from pqeq.in with --pqeqin option. support comment line and flexible number of elements.
• Should save_shell_positions() use MPI I/O to store data into single file? Maybe extend fileio.f90 to support XYZ file format and put the shell position info?
  -> Done. isXYZ dumps atom type and position in xyz format. With isPQEq == .true., it appends the shell charge position too for dielectric property calculation.

[KenichiNomura]

• Current PQEq is very slow. Need to improve the performance. Use Matt's idea; save the intermediate calc during the PQEq iteration.

[KenichiNomura]

Done.