First issue is economics, fuel cost is 1-3% of launch cost. If you can only get half the payload weight to orbit but get most of your rocket back for reuse (and the first stage is the most expensive bit - 9? engines vs one for second stage), cost per Kg to orbit is still (massivly) cheaper.

Second issue is that the fuel cost for the first stage recovery is quite cheap, you only have to brake and land the engine and (almost) empty fuel tank so they are very light vs the lauch mass. From memory a while ago spaceX started using v2 of there main engine which was ~10% more efficent than the v1 engine; This gave enough increased performance that even with extra fuel to land and the extra weight from the landing legs etc. they could get the same payload to orbit plus do styage one recovery.

However this is not a liquid cooling system of the tiles. The (liquid) gas is pumped through the tiles to the leading edge where it is expected to evaporate. So worst case should be no cooling from the gas or the gas layer as a protective layer between the tiles and the incoming atmosphere.

If designed properly if everything works it is re-useable, and if there is a failure you would hope a production model would be designed to that the tiles would survive a single use even without any gas flow.