Add progress output for get_farfields_array

[ChristopherHogan]

Closes #849.
@stevengj @oskooi

[stevengj]

Should be renamed to MEEP_MIN_OUTPUT_TIME if it is in the global header.

[ChristopherHogan]

Fixed.

[ChristopherHogan]

I can't use the time_sink infrastructure to time farfield_lowlevel because am_now_working_on and finished_working are methods of the fields class, while farfield_lowlevel is a dft_near2far method.

[ChristopherHogan]

Added get_farfields_array timing by making am_now_working_on and finished_working public methods and calling them from python.

[oskooi]

The output from running this PR on a test case involving python/examples/binary_grating_n2f.py is:

...
get_farfields_array working on point 470 of 500 (94% done)
get_farfields_array working on point 477 of 500 (95% done)
get_farfields_array working on point 484 of 500 (96% done)
get_farfields_array working on point 491 of 500 (98% done)
get_farfields_array working on point 498 of 500 (99% done)

Elapsed run time = 476.4813 s

Field time usage:
        connecting chunks: 0.113281 s
            time stepping: 120.414 s
            communicating: 4.5503 s
     Fourier transforming: 25.4002 s
        getting farfields: 322.603 s
          everything else: 1.09053 s

It would be useful to include the metric s/point (meaningful for assessing performance) as part of the get_farfields_array working on point ... output similar to s/step for time stepping: on time step 21406 (time=428.12), 0.00672292 s/step. Also, note that the sum of all the sub entries under Field time usage is 474.171311 s. This means that 2.30998 s is still unaccounted for; should this difference just be included in everything else: so that the sum is consistent?

[oskooi]

There is no interval progress output for get_farfields (i.e., get_farfields_array working on point 434 of 500 (86% done)) when using MPI; the only output is the total time at the end of the run:

...
field decay(t = 800.32): 3.1235531072986875e-11 / 0.09814324428153094 = 3.182647089124699e-10
run 0 finished at t = 800.32 (40016 timesteps)

Field time usage:
        connecting chunks: 0.00317784 s
            time stepping: 1.66425 s
            communicating: 0.361284 s
     Fourier transforming: 0.00176502 s
        getting farfields: 323.816 s
          everything else: 1.31586 s