Make mapcoord_threaded faster

[AlexKurek]

I have benchmarked map_2d, as this is where mapcoord_threaded is called from.

before:

   ncalls  tottime  percall  cumtime  percall filename:lineno(function)
       10    0.786    0.079  105.040   10.504 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:584(map_2d)

after:

       10    0.699    0.070   40.033    4.003 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:586(map_2d)
       10    0.707    0.071   39.698    3.970 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:609(map_2d)
       10    0.659    0.066   39.970    3.997 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:604(map_2d)

The entire run on the Xeon 6238R is faster by ~15%. gaul files are identical.

[AlexKurek]

02609bb makes it a bit faster again:

       10    0.668    0.067   35.813    3.581 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:624(map_2d)
       10    0.642    0.064   35.967    3.597 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:624(map_2d)

.gaul files remain identical.

EDIT:
For a larger, 1.5 GB image the entire run is ~9.6% faster:

before: PyBDSF running time: 0:41:0
        8    5.769    0.721  438.971   54.871 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:584(map_2d)
after:  PyBDSF running time: 0:37:4
        8    3.958    0.495  260.992   32.624 /home/akurek/miniconda3/envs/pybdsf-313_local_temp/lib/python3.13/site-packages/bdsf/rmsimage.py:629(map_2d)

.gaul files still remain identical.