Fix incorrect null_count in generate_unions_case integration test

[viirya]

Which issue does this PR close?

Closes #1712.

Rationale for this change

What changes are included in this PR?

Are there any user-facing changes?

[codecov-commenter]

Codecov Report

Merging #1713 (352eed8) into master (20ffaf8) will increase coverage by 0.00%.
The diff coverage is 87.50%.

❗ Current head 352eed8 differs from pull request most recent head 1c206f8. Consider uploading reports for the commit 1c206f8 to get more accurate results

@@           Coverage Diff           @@
##           master    #1713   +/-   ##
=======================================
  Coverage   83.32%   83.32%           
=======================================
  Files         195      195           
  Lines       56012    56018    +6     
=======================================
+ Hits        46671    46677    +6     
  Misses       9341     9341           

Impacted Files	Coverage Δ
arrow/src/ipc/reader.rs	88.72% <80.00%> (-0.08%)	⬇️
arrow/src/ipc/writer.rs	80.37% <100.00%> (+0.06%)	⬆️
parquet_derive/src/parquet_field.rs	65.98% <0.00%> (-0.23%)	⬇️
arrow/src/array/transform/mod.rs	87.02% <0.00%> (+0.22%)	⬆️

---

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 20ffaf8...1c206f8. Read the comment docs.

[tustvold]

LGTM 👍. Just to check my understanding, the bug is that we wrote incorrect null counts, and the logic added to the reader is just a sanity check to make it easier to detect this in future?

[viirya]

@tustvold Yea, that's correct.

[viirya]

Merged. Thanks @tustvold for review.