Benchmark for casting view to dict arrays (and the reverse)

[XiangpengHao]

Which issue does this PR close?

Part of #5861 .

Rationale for this change

Setup a benchmark and see what the performance look like for casting between view and dict arrays.

Related: #5871 #5872

What changes are included in this PR?

Are there any user-facing changes?

[alamb]

Thanks @XiangpengHao -- can you please add this benchmark to the existing cast kernel benchmarks rather than its own bench: https://github.com/apache/arrow-rs/blob/master/arrow/benches/cast_kernels.rs ?

(🤔 maybe we should move that cast_kernels.rs benchmark to arrow-cast/benchmarks )

clicked wrong button

[XiangpengHao]

Thanks @XiangpengHao -- can you please add this benchmark to the existing cast kernel benchmarks rather than its own bench: https://github.com/apache/arrow-rs/blob/master/arrow/benches/cast_kernels.rs ?

(🤔 maybe we should move that cast_kernels.rs benchmark to arrow-cast/benchmarks )

Done.

The cast_kernel.rs can't be moved to arrow-cast/benches because it depends create_primitive_array and seedable_rng which are currently in arrow directory.

[alamb]

Thanks @XiangpengHao

[alamb]

🚀

Merging this in so we can compare performace of main to the specialized cast kernels