feat: Allow casting List<UInt8> to Binary

[itamarst]

Fixes #21549.

[itamarst]

pyodide seems to be failing on main too?

[codecov]

Codecov Report

Attention: Patch coverage is 94.05405% with 11 lines in your changes missing coverage. Please review.

Project coverage is 81.00%. Comparing base (5cedf53) to head (674c82a).
Report is 12 commits behind head on main.

Files with missing lines	Patch %	Lines
crates/polars-compute/src/cast/mod.rs	94.85%	9 Missing ⚠️
crates/polars-core/src/chunked_array/cast.rs	80.00%	2 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##             main   #22611    +/-   ##
========================================
  Coverage   80.99%   81.00%            
========================================
  Files        1661     1664     +3     
  Lines      234867   235162   +295     
  Branches     2772     2773     +1     
========================================
+ Hits       190234   190490   +256     
- Misses      43965    44004    +39     
  Partials      668      668            

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[coastalwhite]

pyodide seems to be failing on main too?

Should be fixed by #22613.

[nameexhaustion]

It should be possible to construct the BinaryViewArray with Views that point directly to the existing Buffer<u8>, instead of copying the memory using an iterator.

There is also a simple optimization we should do where if at the end of construction we see that all Views are in-lined, we can drop the ref to the original buffer.

[itamarst]

@nameexhaustion I did this, and then realized it's a problem insofar as a View assumes a u32 offset but the Buffer<u8> might be bigger than that. Would you happen to know of any prewritten code that splits buffers if they're too large, for this use case?

[itamarst]

@nameexhaustion OK implemented both ideas.

[itamarst]

Thank you for the review, I've addressed the comments.

[itamarst]

Will figure out failing tests tomorrow morning.

[nameexhaustion]

The test is failing because the buffer is now dropped as all the views are in-lined - we should add specific test to assert that there are no data buffers in the result array for the case when all views are inlined.

For the test to ensure proper buffer splitting, the max buffer size limit needs to be increased.