I don't think you could just pre-tension rails so much and be done with it somehow, considering... temperature of a rail can probably stretch between -10 to +50C, and coefficient of expansion for steel is 12*10^-6/K, let delta T = 60K and joined rail length = 1km, the expansion can be +/-36cm(~1.2ft). At some point, the error can be more than one rail segment worth, and maybe it's just me but I can't imagine that amount of expansion just solved by clever tensioning alone.
Usually they are joined to a rail with a slant cut or a tongue section at ends, and let ride into the opposing cuts/tongues, IIUC.
The rail can only expand if it can move. If the rail is clamped down tightly to the sleepers, and the sleepers are tightly packed in the ballast, then it just doesn't expand. The only thing that happens is that stress increases / decreases as temperature changes.
The pre-tensioning is only to ensure that the average tension in rail in the course of an average year is in a certain range.
Usually they are joined to a rail with a slant cut or a tongue section at ends, and let ride into the opposing cuts/tongues, IIUC.