Widely known formulas for radiation power of antennas (e.g. short dipoles) have been derived using Lorentz gauge, which is mathematically convenient, but has no physical justification. I would like to know if these formulas agree at least up to an order of magnitude with the practically observed values.
Use an EM solver, solve a simple antenna and compute the radiated power and compare. EM simulators know nothing of Gauges or retarded potentials...they just crunch down on maxwell's equations or variants thereof, eg Helmholtz equation. Also, since many people have used these for antenna design where radiated power is of utmost concern, there is justification that you can use these for comparison of your analytic results.
that is correct...the lorentz gauge is a simplification of representation of the Vector Potential. However, nearly all high frequency solvers solve in terms of the Electric fields explicitly rather than implicitly through the vector potentials. The exception may be in low frequency analysis where magnetic fields are of primary concern.