The slowdown of testing is not a change in their methodology. It was because they succeeded in all suborbital testing. The orbital testing has been held up by bureaucracy, heat tile iterations, and launch tower refinement.

The costs of test articles are pennies on the dollar compared to the launch value of these vehicles once operational.

By now they have 2.5 stacks ready, and surely have enough raptor engines saved up to rapidly launch 5+ starships this year, assuming they have the heat tile stuff figured out to a degree that the tests would be meaningful.