It's not necessary, but is significantly more radio-quiet than a lunar orbit. And secondly, though unfortunately not something we could really exploit this time, the stable temperatures of the lunar night greatly help with calibration for sensitive measurements like the 21cm Dark Ages signal

Isn’t the benefit here that you don’t have to deal with things such as significant Doppler shift, or having to maintain a supply of propellant for orbit-keeping?

My immediate thought was why not put it in the Earth-Moon L2 Lagrange point, like the James Webb Space Telescope, where it would be permanently shaded from RF from both the Earth and the Sun. But...

1. James Webb is in the Earth-Sun L2 point, where it is largely (though not completely) shaded from the Sun. A radio telescope at Earth-Sun L2 wouldn't be shaded from Earth RF. [edit: JWST is in a halo orbit which keeps it out of the shadow]

2. The Earth-Moon L2 point is shaded from the Earth, but not the Sun. So no benefit compared to the far-side lunar surface.

3. According to TFA, being on the lunar surface gets the telescope out of the solar wind, which is noisy at the low radio frequencies being observed.

There is perhaps some extra opportunity in a 10-14 day solid observation window, but I don't see why a satellite version couldn't still work in smaller windows.

Another reason could be testing for building a much large radio antenna on the moon's surface in the future which is mentioned to farther down in the article. The moon itself and it's dust has electromagnetic effects that might effect measurements and learning about them now could help future planning.

It’s far side, not dark side. The moon doesn’t have a dark side anymore than the earth does