Unfortunately the interactive viewer does not (yet?) have a "click to download..." feature. Your simplest option might be to use the SPARCL server at NOIRLab; see https://astrosparcl.datalab.noirlab.edu and https://github.com/astro-datalab/notebooks-latest/blob/master/04_HowTos/SPARCL/How_to_use_SPARCL.ipynb for info and examples.
For this particular case (and using python), you would install sparcl with
pip install sparclclient
and then get the target with
from sparcl.client import SparclClient
client = SparclClient()
spectra = client.retrieve_by_specid([39628427912810449,],
include=['wavelength', 'flux', 'ivar', 'model', 'wave_sigma'])
spec = spectra.records[0]
# example plotting
import matplotlib.pylab as plt
plt.plot(spec.wavelength, spec.flux)
plt.plot(spec.wavelength, spec.model)
plt.show()
You can use the data in the spec object to save into a fits file, but we don't yet provide an immediate "just give me a fits file for this object" tool. The sparcl notebook above also has examples for how to query by other parameters like ra,dec if you don't already know the targetid.
For the spectra metadata, the only way I've found to discover the options is to request something that doesn't exist (like "lsf") and then it will print a helpful error message with what you can request:
BadInclude: [BADINCL] The INCLUDE list (flux,ivar,lsf,model,wavelength) contains invalid data field names for Data Sets (BOSS-DR16,DESI-EDR,SDSS-DR16). Unknown fields are: lsf. Available fields are: data_release, datasetgroup, dateobs, dateobs_center, dec, dirpath, exptime, extra_files, filename, filesize, flux, instrument, ivar, mask, model, ra, redshift, redshift_err, redshift_warning, site, sparcl_id, specid, specprimary, spectype, survey, targetid, telescope, updated, wave_sigma, wavelength, wavemax, wavemin.
Other options