How would they compare to regular sodium bulbs? Due to its high vapor pressure, it could work well. Also, potassium could be an option.

I think rubidium would yield lower efficacy than sodium. That why caesium and potassium lamps weren't existed commercially, but only as engineering samples.

I'd like to see a lithium filled HPS (HPL?) lamp just since lithium is so pretty when it's energized.

Lithium does not work in an HPS lamp, I have tried it without success!  The vapour pressure of this metal is just too low.  However Rubidium, Potassium and Caesium do all work, but have terribly poor luminous efficacy.  At very high loadings under pulse operation its possible to achieve about 40 lm/W, but standard HPS type lamps with caesium are only about half as efficient as incandescent lamps of the same rating.  If I remember rightly the best I ever achieved in a 600W caesium lamp was about 12 lm/W!  Beautiful colour white light of very high CRI though.

Lithium does not work in an HPS lamp, I have tried it without success!  The vapour pressure of this metal is just too low.  However Rubidium, Potassium and Caesium do all work, but have terribly poor luminous efficacy.  At very high loadings under pulse operation its possible to achieve about 40 lm/W, but standard HPS type lamps with caesium are only about half as efficient as incandescent lamps of the same rating.  If I remember rightly the best I ever achieved in a 600W caesium lamp was about 12 lm/W!  Beautiful colour white light of very high CRI though.

Isn't that result of a fact the arc being then more of a a Planck black body type than a quantum excitation-emission type radiator?