I get your point, but I still don't see that it will matter. If you read their testing methodology they are really only disabling cache to simulate real world usage of going to say 25 (or hundreds) different web pages for the first time. They just disable cache because they are reloading the same page over and over again from their local test server hosting that page.
The test is to gauge how long laptop batteries last. If they turn on caches for all laptops and run the same test, that would be OK too. If they let you or me pick 25 web pages to load over and over, that would be OK too. If they raise the number to 50 web pages or 500 web pages, that's fine too as long as the test is the same for all of the laptops they test.
The point is that it's just a standard testing protocol they apply to
ALL of their laptop reviews. The specifics of it are almost irrelevant as long as whatever protocols they want to use are consistently applied to the many laptops they test.
In this story, Apple interacted with them, found a software bug that contributed to the battery burn and it sounds like CR is willing to have another go at a fixed MBpro. However, some of the ongoing spin seems to revolve around the idea of how not using the cache is unnatural because using it is default... ignoring that the same standard is being applied to all of the other laptops they test too. Or more simply, great effort being made to fault CR or redirect the underlying issue away from Apple.
Use the cache or not, but the point is that if the protocol is going to change, it can't change for the just this one laptop... so that just it can perform better. It's not CR's business to help sell one corporations laptop. It's CR, not PR. And I'm not CR, nor even expert at what CR does, but I'll go ahead and concede with great confidence that that particular protocol change- if applied- will defiantly improve the battery life result for that test. CR themselves says so in the part of their quote where they explain whey they turn it off for their tests.