Fix Particle file output spacing for consistency and ease of parsing

[ken-lauer]

Changes

1. Fix leading/trailing spaces in particle file output
2. Add compilation documentation when using conda-forge's compilers
3. Fix CI building/running

Particle file output

This fixes the leading spaces in output particle files. This saves a bit of space and makes it slightly easier for not-so-smart parsers to read the output particles (specifically polars.read_csv on our end, which can be ~10x faster than numpy.loadtxt for large particle output files).

Currently particles look like this:

   0.192544220656E-03  -0.451056940538E-04   0.472496073462E-08   0.564525641813E-03  -0.107436741277E-05   0.115771507881E-02  -0.195692801440E-05  -0.000000000000E+00   0.100000000000E+01

After this PR, the leading space and double spaces between should be removed. g0 formatting looks like the following:

0.19254422065579266E-3 -0.45105694053811659E-4 0.47249607346183573E-8 0.56452564181341889E-3 -0.10743674127672016E-5 0.11577150788125525E-2 -0.19569280144029904E-5 -0.0000000000000000 1.0000000000000000

Fortunately, there does not appear to be a trailing space inserted with 9(g0,1x).

CI fix

• A necessary flag for the standard GNU make workflow to work with modern compilers wasn't being passed (-fallow-argument-mismatch)
• MPICH on GitHub Actions appears to have issues assigning the number of requested processors for the communicator
  • This made running the first example with 2+ cores impossible. (See notes below)
  • It would result in the error: MPIR_Cart_create_impl(41): Size of the communicator (1) is smaller than the size of the Cartesian topology (2)
  • I adjusted MPICH to only run 1x1 processes, to at least verify that the build worked.

(Similar to impact-lbl/IMPACT-Z#12)