Change Block getSizeInBytes to estimate fully expanded data size

[dain]

Description

The Block getSizeinBytes method has been defined to return the "compacted" size of a block. This means for an RLE only the size of the single value is returned, and for a dictionary block the size of each position is computed from the dictionary. This computation is quite expensive and requires Blocks to have getPositionsSizeInBytes and fixedSizeInBytesPerPosition methods.

This PR changes the implementation to be defined as returning an estimate of the full data size of a block. This means for an RLE block the size is value.getSizeInBytes() * positionCount and dictionary is (dictionary.getSizeInBytes() / dictionary.getPositionCount()) * positionCount, which is simpler and much faster to compute.

Current Usage

The getSizeinBytes method is typically used to estimate the size needed for output buffers when copying data in operators and generally these usages assume that the value returned is a fully expanded size. The other main usage is in stats, and here the change will have visible effects. This method is often used to calculate the input/output size of operators, sources, and sinks. It is not clear if the these uses intended to have compacted or fully expanded size.

Release notes

(X) This is not user-visible or is docs only, and no release notes are required.
( ) Release notes are required. Please propose a release note for me.
( ) Release notes are required, with the following suggested text:

## SPI
* Block `getSizeInBytes` now returns an estimate of the full data size of the block. ({issue}`issuenumber`)