Optimize Shuffle.decode method

[aaime]

Description of Changes

While doing a profiling session I've noticed this method popping up over and over. Noticed there were some simple and obvious optimizations, like not allocating an extra array, and avoiding repeated multiplications in index computation, and thought to share the results with the wider community.

PR Checklist

• [ ] Link to any issues that the PR addresses
• [ ] Add labels
• [ ] Open as a draft PR until ready for review
• [ ] Make sure GitHub tests pass
• [ ] Mark PR as "Ready for Review"

[CLAassistant]

All committers have signed the CLA.