merge files fails with small or large files

Sorry if there is already an issue for this. the word "merge" is difficult to search for due to its use in git.

The merge function throws an error if any parts are outside the valid part size range for an s3 multipart upload.

Large input files can become multiple parts using the CopySourceRange parameter of upload_part_copy
Smaller parts will either need at least 5MiB of data to be downloaded or temporary files on s3.

Here are some failing tests

def test_merge_small_part(s3):
    with s3.open(a, "wb") as f:
        f.write(b"a" * 4 * 2**20)

    with s3.open(b, "wb") as f:
        f.write(b"a" * 10 * 2**20)
    s3.merge(test_bucket_name + "/joined", [a, b])
    assert s3.info(test_bucket_name + "/joined")["size"] == (4 * 2**20) + (10 * 2**20)

def test_merge_large_part(s3):
    with s3.open(a, "wb") as f:
        f.write(b"a" * 6 * 2**30)

    with s3.open(b, "wb") as f:
        f.write(b"a" * 10 * 2**20)
    s3.merge(test_bucket_name + "/joined", [a, b])
    assert s3.info(test_bucket_name + "/joined")["size"] == (6 * 2**30) + (10 * 2**20)

Start at the s3.merge entry point and inspect how it uses upload_part_copy and the CopySourceRange parameter. Reproduce the two failing cases shown in the issue, covering both a small and a large part. Done means both tests complete successfully and the joined object size matches the sum of the inputs.