chore: Add join benchmarks

[coderfender]

Which issue does this PR close?

We don't currently have benchmarks for just the join implementation. This PR essentially bridges that gap

Closes #.

Rationale for this change

What changes are included in this PR?

How are these changes tested?

[coderfender]

Running benchmark: inner count
  Running case: Spark
  Stopped after 20 iterations, 2043 ms
  Running case: Comet
  Stopped after 34 iterations, 2002 ms

OpenJDK 64-Bit Server VM 17.0.16+8-LTS on Mac OS X 26.3.2
Apple M5 Pro
inner count:                              Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
------------------------------------------------------------------------------------------------------------------------
Spark                                                89            102           9         11.7          85.2       1.0X
Comet                                                51             59           6         20.6          48.5       1.8X

Running benchmark: inner projected
  Running case: Spark
  Stopped after 28 iterations, 2021 ms
  Running case: Comet
  Stopped after 48 iterations, 2029 ms

OpenJDK 64-Bit Server VM 17.0.16+8-LTS on Mac OS X 26.3.2
Apple M5 Pro
inner projected:                          Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
------------------------------------------------------------------------------------------------------------------------
Spark                                                66             72           5         15.9          62.7       1.0X
Comet                                                36             42           3         29.1          34.4       1.8X

Running benchmark: left outer
  Running case: Spark
  Stopped after 27 iterations, 2044 ms
  Running case: Comet
  Stopped after 43 iterations, 2039 ms

OpenJDK 64-Bit Server VM 17.0.16+8-LTS on Mac OS X 26.3.2
Apple M5 Pro
left outer:                               Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
------------------------------------------------------------------------------------------------------------------------
Spark                                                68             76           6         15.4          64.8       1.0X
Comet                                                40             47           4         26.2          38.1       1.7X

Running benchmark: left semi
  Running case: Spark
  Stopped after 26 iterations, 2036 ms
  Running case: Comet
  Stopped after 43 iterations, 2023 ms

OpenJDK 64-Bit Server VM 17.0.16+8-LTS on Mac OS X 26.3.2
Apple M5 Pro
left semi:                                Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
------------------------------------------------------------------------------------------------------------------------
Spark                                                67             78           9         15.8          63.4       1.0X
Comet                                                39             47           5         27.1          36.8       1.7X

Running benchmark: right outer
  Running case: Spark
  Stopped after 8 iterations, 2152 ms
  Running case: Comet
  Stopped after 25 iterations, 2040 ms

OpenJDK 64-Bit Server VM 17.0.16+8-LTS on Mac OS X 26.3.2
Apple M5 Pro
right outer:                              Best Time(ms)   Avg Time(ms)   Stdev(ms)    Rate(M/s)   Per Row(ns)   Relative
------------------------------------------------------------------------------------------------------------------------
Spark                                               264            269           4          4.0         252.2       1.0X
Comet                                                74             82           4         14.2          70.4       3.6X