fix: too large data chunk generated by highly compressed yet nested data with RLE

[Xuanwo]

close #4429

---

As described in #4429, highly compressed yet nested data using RLE can produce data chunks that exceed our 16KiB threshold. This happens because our RLE encoding currently considers only the data buffer size and does not account for the size of REP/DEF markers, which can consume up to 4 bytes per value.

Ideally, we should include REP/DEF sizes in the calculation, but that would require significant changes. In this PR, I implemented a workaround to address the issue at the cost of a slightly lower compression ratio. A more comprehensive fix will follow after discussion.

This PR also includes a repro as part of our unit test to prevent regression of this bug.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 81.90%. Comparing base (0c69144) to head (0c758c7).
⚠️ Report is 2 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4431      +/-   ##
==========================================
+ Coverage   81.88%   81.90%   +0.02%     
==========================================
  Files         302      302              
  Lines      123146   123298     +152     
  Branches   123146   123298     +152     
==========================================
+ Hits       100839   100990     +151     
- Misses      18502    18506       +4     
+ Partials     3805     3802       -3     

Flag	Coverage Δ
unittests	81.90% <100.00%> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.