Please include any website links that answer this question.
Not "bombard," it's mostly just heating; heating the glass surface, and greatly heating the metal electrodes. Not "impurities," it's mostly just water.
The answer will be found on all the vac. tech sites about "bake-out" of vacuum chambers in professional research. (The neon signmaker sites seem all to treat the bombard step as a mystery.) E.g. see WP bake-out
The surface of normal, undessicated glass is full of adsorbed H2O. If you try to pump a vacuum in a glass chamber, the surface water will desorb, and your vacuum will slowly fill with water vapor. If plasma etches the glass surface, as with neon signs, the water can escape the glass more quickly, and your vac gauge will show a jump in pressure (and rapid color change) when the sign first turns on.
We can avoid filling our neon signs with hydrogen and oxygen by initially heating the glass surface to well above 100C boiling point, while running the vacuum pumps to eject the evaporating water vapor. (Ideally use a cold trap to grab all the water, so it doesn't ruin your pump oil.)
"Burning the impurities" might apply to the metal electrodes, since they very well may be covered with fingerprints which contain H2O, volatile oils, and significant gas. Clean the metal parts first (acetone. The pros use ultrasonic acetone bath.) Once installed, you'll want to heat the metal electrodes beyond red hot, to char then evaporate any remaining traces of finger-slime left on them by the manufacturer. To do this, also the adjacent glass must be extremely hot at the time, to avoid condensing any soot etc. on a cold glass surface.
And last, if you're using a cheap oil-vane vacuum pump without coldtrap and filters, then the pump may spew oil-mist into your vacuum, which settles on all the inner surfaces.
Without needing any giant glass annealing kiln to heat your whole neon sign, the electrical "bombarding" step will heat the metal electrodes to evaporate unwanted surface material, boil off all the oil micro-droplets, and heat the glass tube to well above 500F to outgas all the adsorbed water.
Finally, if the gas in your vacuum system is heated hot enough to glow, it will have an unusually low density, and when cooled down again, will give a better vacuum. The "bombard" step takes gas out of the solid surfaces, but it also may take gas out of the empty space itself, giving a better vacuum than your mechanical pump can give. (Can't do that with an annealing oven; not without melting the glass.)
Search term: chamber bakeout