Base.summarysize for Memory with Union

[PatrickHaecker]

Fixes the double accounting of the union byte array in Base.summarysize as described in #57506.

If this is the correct fix, can it be backported to 1.11?

Fix #57506

[PatrickHaecker]

This is only one possible fix. The alternative would be to not account for the union byte array in sizeof. This depends on the exact definition of sizeof which is discussed in #54007.

It might be that there is a more consistent way to define sizeof. If this is done, the implementation of Base.summarysize might need to be adapted. However, the current result of Base.summarysize is just wrong and should therefore be fixed.

[LilithHafner]

Thanks for contributing, @PatrickHaecker! This looks great to me. I'm going to wait a bit before merging to see of @oscardssmith wants to chime in.

[oscardssmith]

thanks for this!