PoC Adaptive round robin repartitioning

[Dandandan]

Which issue does this PR close?

TODO

• Verify results, check if the perf. change comes from adaptive repartitioning or from other changes
• Run more benchmarks
• Try to be more fair (e.g. don't start at 0)
• Introduce more round robin repartitioning for sources (even when already splitted)
• Put it behind config (on by default?)
• Fix / improve test

--------------------
Benchmark tpch_mem_sf1.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query        ┃     main ┃ repartition_exec_poc ┃        Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1     │  83.38ms │              76.98ms │ +1.08x faster │
│ QQuery 2     │  15.09ms │              13.93ms │ +1.08x faster │
│ QQuery 3     │  25.32ms │              23.96ms │ +1.06x faster │
│ QQuery 4     │  13.39ms │              13.49ms │     no change │
│ QQuery 5     │  40.86ms │              38.85ms │     no change │
│ QQuery 6     │   6.18ms │               5.23ms │ +1.18x faster │
│ QQuery 7     │  81.24ms │              76.82ms │ +1.06x faster │
│ QQuery 8     │  19.83ms │              19.37ms │     no change │
│ QQuery 9     │  49.55ms │              44.83ms │ +1.11x faster │
│ QQuery 10    │  41.14ms │              37.67ms │ +1.09x faster │
│ QQuery 11    │   7.19ms │               6.23ms │ +1.15x faster │
│ QQuery 12    │  29.19ms │              26.92ms │ +1.08x faster │
│ QQuery 13    │  21.48ms │              18.15ms │ +1.18x faster │
│ QQuery 14    │   6.01ms │               5.88ms │     no change │
│ QQuery 15    │  13.09ms │              15.43ms │  1.18x slower │
│ QQuery 16    │  14.31ms │              14.30ms │     no change │
│ QQuery 17    │  55.49ms │              62.14ms │  1.12x slower │
│ QQuery 18    │ 151.61ms │             150.71ms │     no change │
│ QQuery 19    │  28.32ms │              24.11ms │ +1.17x faster │
│ QQuery 20    │  24.71ms │              21.99ms │ +1.12x faster │
│ QQuery 21    │ 111.26ms │             103.85ms │ +1.07x faster │
│ QQuery 22    │  14.59ms │              13.03ms │ +1.12x faster │
└──────────────┴──────────┴──────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━┓
┃ Benchmark Summary                   ┃          ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━┩
│ Total Time (main)                   │ 853.23ms │
│ Total Time (repartition_exec_poc)   │ 813.85ms │
│ Average Time (main)                 │  38.78ms │
│ Average Time (repartition_exec_poc) │  36.99ms │
│ Queries Faster                      │       14 │
│ Queries Slower                      │        2 │
│ Queries with No Change              │        6 │
└─────────────────────────────────────┴──────────┘
Note: Skipping /Users/danielheres/Code/arrow-datafusion-personal/benchmarks/results/main/tpch_mem_sf10.json as /Users/danielheres/Code/arrow-datafusion-personal/benchmarks/results/repartition_exec_poc/tpch_mem_sf10.json does not exist
--------------------
Benchmark tpch_sf1.json
--------------------
┏━━━━━━━━━━━━━━┳━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━┓
┃ Query        ┃     main ┃ repartition_exec_poc ┃        Change ┃
┡━━━━━━━━━━━━━━╇━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━┩
│ QQuery 1     │ 125.52ms │             104.10ms │ +1.21x faster │
│ QQuery 2     │  57.14ms │              36.75ms │ +1.56x faster │
│ QQuery 3     │  56.00ms │              55.02ms │     no change │
│ QQuery 4     │  41.75ms │              40.46ms │     no change │
│ QQuery 5     │  78.27ms │              73.82ms │ +1.06x faster │
│ QQuery 6     │  21.36ms │              21.31ms │     no change │
│ QQuery 7     │  93.01ms │              95.74ms │     no change │
│ QQuery 8     │  79.40ms │              75.75ms │     no change │
│ QQuery 9     │ 115.29ms │             100.02ms │ +1.15x faster │
│ QQuery 10    │ 101.89ms │             100.49ms │     no change │
│ QQuery 11    │  40.69ms │              26.26ms │ +1.55x faster │
│ QQuery 12    │  54.58ms │              50.62ms │ +1.08x faster │
│ QQuery 13    │ 122.04ms │              80.41ms │ +1.52x faster │
│ QQuery 14    │  39.81ms │              40.11ms │     no change │
│ QQuery 15    │  49.66ms │              51.29ms │     no change │
│ QQuery 16    │  36.18ms │              25.92ms │ +1.40x faster │
│ QQuery 17    │ 105.28ms │              94.82ms │ +1.11x faster │
│ QQuery 18    │ 164.32ms │             143.05ms │ +1.15x faster │
│ QQuery 19    │  65.62ms │              58.38ms │ +1.12x faster │
│ QQuery 20    │  60.52ms │              63.31ms │     no change │
│ QQuery 21    │ 130.67ms │             120.20ms │ +1.09x faster │
│ QQuery 22    │  35.66ms │              28.42ms │ +1.25x faster │
└──────────────┴──────────┴──────────────────────┴───────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━┓
┃ Benchmark Summary                   ┃           ┃
┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━┩
│ Total Time (main)                   │ 1674.66ms │
│ Total Time (repartition_exec_poc)   │ 1486.23ms │
│ Average Time (main)                 │   76.12ms │
│ Average Time (repartition_exec_poc) │   67.56ms │
│ Queries Faster                      │        13 │
│ Queries Slower                      │         0 │
│ Queries with No Change              │         9 │
└─────────────────────────────────────┴───────────┘

Closes #.

Rationale for this change

In
#7001

What changes are included in this PR?

Are these changes tested?

Are there any user-facing changes?