perf: Add dedicated concatenate kernels

[orlp]

This mainly speeds up our rechunks, previously this went through the generic make_growable interface which isn't as efficient with getting the allocation sizes correct, and was also unnecessarily copying string data for ViewArrays. EDIT: doing lazy string copying slowed down the new-streaming engine due to an excessive amount of buffers in the string view, so that's disabled for now.

[codecov]

Codecov Report

Attention: Patch coverage is 83.06011% with 62 lines in your changes missing coverage. Please review.

Project coverage is 79.25%. Comparing base (3336475) to head (afe55b8).
Report is 46 commits behind head on main.

Files with missing lines	Patch %	Lines
crates/polars-arrow/src/compute/concatenate.rs	84.19%	55 Missing ⚠️
crates/polars-arrow/src/array/values.rs	0.00%	2 Missing ⚠️
crates/polars-core/src/chunked_array/from.rs	0.00%	2 Missing ⚠️
crates/polars-arrow/src/legacy/array/list.rs	0.00%	1 Missing ⚠️
crates/polars-core/src/series/from.rs	0.00%	1 Missing ⚠️
crates/polars-json/src/ndjson/deserialize.rs	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #21080      +/-   ##
==========================================
+ Coverage   79.20%   79.25%   +0.05%     
==========================================
  Files        1586     1585       -1     
  Lines      225648   225985     +337     
  Branches     2588     2588              
==========================================
+ Hits       178714   179100     +386     
+ Misses      46344    46295      -49     
  Partials      590      590              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.