Fix CSV COPY to S3 using FILE_SIZE_BYTES

[BrynjarM]

FILE_SIZE_BYTES did not influence CSV COPY operations to S3 since, unlike the filesystem variant, the file length was consistently reported as 0. The implementation now updates the length as data is written + test added (based on https://github.com/duckdb/duckdb/blob/main/test/sql/copy/file_size_bytes.test).

Fixes duckdb/duckdb#18381

[BrynjarM]

@Tmonster, @carlopi, @samansmink
I'm bringing this pull request to your attention as I'm unable to formally request reviewers. I noticed you are all active contributors and was hoping you might have a moment to look.
My primary goal with this PR was to clearly illustrate the issue from the bug report (that I found after running into this exact issue). As I'm new to the codebase, please feel free to take a completely different approach to the solution.
The key outcome is to ensure the FILE_SIZE_BYTES setting is correctly respected when writing CSVs to S3, allowing us to control the number of output files.
Thank you for your time and consideration.