This example shows how to use ph.x on a GRID. The GRID phonon parallelization
allows to split a phonon calculation into relatively small jobs, each
runing a subset of linear-response calculations for a small number of
wave-vectors or irreps or compbinations. The jobs can then be run
independently and the final results collected. Documentation is available
here: :Calculation of Phonon Dispersions on the GRID using Quantum ESPRESSO",
R. di Meo, A. Dal Corso, P. Giannozzi, and S. Cozzini, in
"Chemistry and Material Science Applications on Grid Infrastructures", Editors:
S. Cozzini, A. Lagana', ICTP Lecture Notes Series, Vol.24, pp.165-183 (2009)
http://users.ictp.it/~pub_off/lectures/lns024/10-giannozzi/10-giannozzi.pdf

The calculation is the same as example06. In run_example all q-vectors and
all irreps are split into different runs. In run_example_1 all q-vectors are 
split into different runs.

The calculation in run_example proceeds as follows:

1) A self consistent calculation is done:
(input alas.scf.in, output alas.scf.out)

2) A preparatory phonon run with start_irr=0, last_irr=0 calculates the
displacement patterns:
(input alas.ph.in0, output alas.ph.out0)

3) ph.x is run for each representation of each q point. 
The code runs with different outdir and only the xml files are copied 
in the same outdir
(input input.#q.#irr, output output.#q.#irr)

4) A final phonon calculation collects all the dynamical matrices
 (input alas.ph.in, output alas.ph.out)

5) q2r and matdyn compute the interatomic force constants and the dispersions.

The calculation in run_example_1 proceeds as follows:

1) A self consistent calculation is done:
(input alas.scf.in, output alas.scf.out)

2) ph.x is run for each q point. 
The code runs with different outdir. This step can be done in different
machines or different processors. 
(input input.#q, output output.#q)

3) q2r and matdyn compute the interatomic force constants and the dispersions.
