Benchmark Spilling

[quasiben]

We should build a benchmark to profile spilling in at least the following scenarios:

• performance: measure that our spilling is efficient in a larger workflow -- maybe this is the merge benchmark with larger than memory data
• stability: ensure that we never OOM

We should also test these scenarios with both the exist spilling infrastructure and the RAPIDS Spilling Manager

Briefly discussed with @wence- and @charlesbluca

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.