Seems like there is confusion around different testing processes. In the software world we have unit, system, integration, usability, user acceptance testing (to name a few). All of this occurs on the same software but with somewhat different goals. Seems like a similar situation with pinball (different types of testing). Prototypes are built in order to validate (test) and improve the efficacy of the design. Production testing is done to validate (test) and improve the efficacy of the production process. This requires the machines be 'put through its paces' in order to validate the production process results in a solid product. How long they will undergo this testing is probably more a function of what is discovered than anything else (although I assume it is initially time-boxed)