Improve performance if dictionary kernels, add benchmark and add take_iter_unchecked

[viirya]

Which issue does this PR close?

Closes #1371.

dict eq string          time:   [682.94 us 686.12 us 689.61 us]                                                                                                   
                        change: [-3.5494% -2.4463% -1.3917%] (p = 0.00 < 0.05)                                                                                    
                        Performance has improved.                                                                                                                 
Found 9 outliers among 100 measurements (9.00%)                                                                                                                   
  6 (6.00%) high mild                                                                                                                                             
  3 (3.00%) high severe                                                                                                                                           

Rationale for this change

What changes are included in this PR?

Are there any user-facing changes?

[codecov-commenter]

Codecov Report

Merging #1372 (4f41726) into master (6ee51bc) will increase coverage by 0.03%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master    #1372      +/-   ##
==========================================
+ Coverage   82.99%   83.02%   +0.03%     
==========================================
  Files         181      181              
  Lines       52933    53020      +87     
==========================================
+ Hits        43932    44022      +90     
+ Misses       9001     8998       -3     

Impacted Files	Coverage Δ
arrow/src/compute/kernels/comparison.rs	92.45% <ø> (+0.29%)	⬆️
arrow/src/array/array_binary.rs	93.22% <100.00%> (-0.27%)	⬇️
arrow/src/array/array_boolean.rs	93.18% <100.00%> (-1.44%)	⬇️
arrow/src/array/array_primitive.rs	94.71% <100.00%> (+0.02%)	⬆️
arrow/src/array/array_string.rs	97.65% <100.00%> (+0.01%)	⬆️
arrow/src/datatypes/datatype.rs	66.40% <0.00%> (-0.40%)	⬇️
parquet_derive/src/parquet_field.rs	65.75% <0.00%> (-0.23%)	⬇️
arrow/src/compute/kernels/temporal.rs	97.22% <0.00%> (-0.13%)	⬇️
... and 4 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6ee51bc...4f41726. Read the comment docs.

[viirya]

The performance looks a bit better?

cc @alamb

[alamb]

Thank you @viirya -- I think this looks 👍

[viirya]

Thank you @alamb !

[alamb]

Thank you @viirya -- sorry I am backed up on reviews and merging. Always so much to do!

[viirya]

Thank you @alamb again!