fix: text partitioning logic

[piksel]

The fix in #240 only addressed a specific issue, regarding repeated sequences of strings with no whitespace that were exact multiples of the maximum allow chunk size. It just moved the problem to another offset instead.

This does instead fix the real underlying issue and fixing one oddity with previous implementation (why a text string that were the same length as the chunk size would not fit inside one chunk).

Additionally, this includes a test iterating through 8000 text lengths and verifying the contents as well as the chunk sizes.
The general problem with the test is that the logic for calculating the expected sizes works the same way as the implementation. But matched with the unaligned limits and iteration, it should verify the logic sufficiently.

Fixes #244.

[codecov]

Codecov Report

Merging #245 (c12a45e) into main (bafa90a) will decrease coverage by 0.00%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #245      +/-   ##
==========================================
- Coverage   77.55%   77.54%   -0.01%     
==========================================
  Files          95       95              
  Lines        2860     2859       -1     
==========================================
- Hits         2218     2217       -1     
  Misses        469      469              
  Partials      173      173              

Impacted Files	Coverage Δ
pkg/util/util.go	100.00% <ø> (ø)
pkg/util/partition_message.go	90.00% <100.00%> (ø)

---

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 bafa90a...c12a45e. Read the comment docs.

[piksel]

There is something odd going on with the coverage. There is one less covered line, but that line has been removed in this PR, so it shouldn't count.