Make rapids-download-wheels-from-s3 work like rapids-download-conda-from-s3

[ajschmidt8]

The rapids-download-wheels-from-s3 script works differently from rapids-download-conda-from-s3 and that has made it difficult to refactor some things in #59.

Their functionality differs in that the conda script extracts the downloads to an arbitrary directory and prints the path whereas the wheel script accepts a path that it should extract to and doesn't print anything.

• conda example - https://github.com/rapidsai/cudf/blob/d4c2d1ccee07005a02907d0e99a1057ec813c084/ci/test_python_common.sh#L24
• wheel example - https://github.com/rapidsai/cibuildwheel-imgs/blob/d4a61bfda206ab37ec7cad19dd0c122d013bd11d/citestwheel/citestwheel.sh#L23

For the sake of consistency, it would be nice if these two scripts behaved the same. Additionally, if their implementation was consistent, we could likely remove tools/_rapids-download-from-s3 and move its logic to tools/rapids-download-from-s3 (I forget the exact details, but I think that's what I was trying to achieve in #59).

[ajschmidt8]

Linking this comment for additional context: rapidsai/cuxfilter#506 (comment)