I think the resistance is that we don't really want to incentivize spending engineering on performance optimizations for a path that we actively discourage authors from using.

We likely want to track the willReadFrequently:false testcase individually in our own regression testing, but I'd prefer not to encourage engineering investment in something that, should an author wish to, is loudly warned about and easily fixed.

I think we're not totally against having tests for both, but that I feel like we'd prefer to weight them as something like 30:1 for willReadFrequently:true vs false. In practice, this is akin to not having the willReadFrequently:false/default case.