perf: Perform Local Distinct list_agg in count_distinct aggregations

[srilman]

Changes Made

Currently, most of the actual work for count_distinct is done at the finalization step, which only runs on 1 task / thread. Thus, very little parallelism is possible. With this change, Daft will perform a local distinct operation on micro-partitions during the sink phase, which can be done in parallel.

Performance results of running ClickBench Queries 4 & 5 (6 runs each, skipping first) on a M4 Max MacBook Pro w/ 14 cores & 32 GB memory:

Query	Before	After	Improvement
Q4	2.95s	1.89s	1.56x
Q5	2.82s	1.30s	2.16x

Related Issues

Checklist

• Documented in API Docs (if applicable)
• Documented in User Guide (if applicable)
• If adding a new documentation page, doc is added to docs/mkdocs.yml navigation
• Documentation builds and is formatted properly (tag @/ccmao1130 for docs review)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.27%. Comparing base (5117945) to head (527a080).
Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #4325   +/-   ##
=======================================
  Coverage   78.27%   78.27%           
=======================================
  Files         818      818           
  Lines      107999   107999           
=======================================
+ Hits        84531    84532    +1     
+ Misses      23468    23467    -1     

Files with missing lines	Coverage Δ
...ft-physical-plan/src/physical_planner/translate.rs	93.45% <100.00%> (ø)

... and 2 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.