Store velocity vectors in a reservoir

[lstagner]

This PR implements storing the neutral particle velocity vectors that pass through each cell. When calculating cx rates the cross sections are averaged over the velocities passing through the cell for each species.

This change makes it easier to support multiple beams/beamlets, calculating FIDA-halo, and mixed species support.

This is a breaking PR as it changes the neutrals.h5 file format to accommodate the particle reservoir. The file is now as follows

[runid]_neutrals.h5
    |-full
        |- amu # mass of the neutral species
        |- nlevel # number of energy levels in neutrals
        |- dens # neutral density
        |- reservoir
            |- v    # neutral particle velocity (3, k, nx, ny ,nz)
            |- w   # neutral particle weight (k, nx, ny, nz)
            |- n    # Number of neutrals seen by the cell (nx,ny,nz)
            |- k    # size of the reservoir
    |-half
...

[lstagner]

I did a comparison for a D3D Main Ion CER case. Biggest difference is that the new reservoir sampling sees more low energy photons in the halo. @shaunhaskey

DCX:

HALO:

[shaunhaskey]

@lstagner I fit the output spectra with a Gaussian like I usually do for calculating the apparent main-ion signals. As expected based on the plots from @lstagner, there is minimal change between the versions.