fix: Fix zstd decompression of multi-frame responses

[josumoreno-BP]

ZstdDecompressor is one-shot-per-frame: once a frame ends, subsequent decompress() calls raise EOFError. This broke HTTP responses where the server sends multiple zstd frames (common with chunked transfer encoding).

Detect frame boundaries via eof/unused_data attributes and create fresh decompressor instances for subsequent frames.

What do these changes do?

Detect frame boundaries via eof/unused_data attributes and create fresh decompressor instances for subsequent frames.

Are there changes in behavior for the user?

The change is seamless for the uses

Is it a substantial burden for the maintainers to support this?

I think this is needed to cover multi-framed zstd responses. I tried to keep it simple but covering the cases that came to my mind.

Related issue number

Fixes #12234

Checklist

• I think the code is well written
• Unit tests for the changes exist
• Documentation reflects the changes - I don't think we need to update the docs in this case
• If you provide code modification, please add yourself to CONTRIBUTORS.txt
  • The format is <Name> <Surname>.
  • Please keep alphabetical order, the file is sorted by names.
• Add a new news fragment into the CHANGES/ folder

  • name it <issue_or_pr_num>.<type>.rst (e.g. 588.bugfix.rst)

  • if you don't have an issue number, change it to the pull request
    number after creating the PR

    • .bugfix: A bug fix for something the maintainers deemed an
      improper undesired behavior that got corrected to match
      pre-agreed expectations.
    • .feature: A new behavior, public APIs. That sort of stuff.
    • .deprecation: A declaration of future API removals and breaking
      changes in behavior.
    • .breaking: When something public is removed in a breaking way.
      Could be deprecated in an earlier release.
    • .doc: Notable updates to the documentation structure or build
      process.
    • .packaging: Notes for downstreams about unobvious side effects
      and tooling. Changes in the test invocation considerations and
      runtime assumptions.
    • .contrib: Stuff that affects the contributor experience. e.g.
      Running tests, building the docs, setting up the development
      environment.
    • .misc: Changes that are hard to assign to any of the above
      categories.

  • Make sure to use full sentences with correct case and punctuation,
    for example:

    Fixed issue with non-ascii contents in doctest text files
    -- by :user:`contributor-gh-handle`.

    Use the past tense or the present tense a non-imperative mood,
    referring to what's changed compared to the last released version
    of this project.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.11%. Comparing base (993989c) to head (9a000fd).
⚠️ Report is 145 commits behind head on master.
✅ All tests successful. No failed tests found.

Additional details and impacted files

@@           Coverage Diff           @@
##           master   #12290   +/-   ##
=======================================
  Coverage   99.11%   99.11%           
=======================================
  Files         130      130           
  Lines       45446    45524   +78     
  Branches     2398     2403    +5     
=======================================
+ Hits        45043    45122   +79     
  Misses        272      272           
+ Partials      131      130    -1     

Flag	Coverage Δ
CI-GHA	98.97% <98.75%> (+<0.01%)	⬆️
OS-Linux	98.71% <98.75%> (ø)
OS-Windows	96.98% <98.75%> (+<0.01%)	⬆️
OS-macOS	97.87% <98.75%> (+<0.01%)	⬆️
Py-3.10.11	97.42% <97.50%> (+<0.01%)	⬆️
Py-3.10.20	97.89% <97.50%> (-0.01%)	⬇️
Py-3.11.15	98.09% <97.50%> (-0.01%)	⬇️
Py-3.11.9	97.63% <97.50%> (-0.01%)	⬇️
Py-3.12.10	97.72% <97.50%> (+<0.01%)	⬆️
Py-3.12.13	98.19% <97.50%> (-0.01%)	⬇️
Py-3.13.12	98.43% <97.50%> (-0.01%)	⬇️
Py-3.14.3	98.49% <97.50%> (-0.01%)	⬇️
Py-3.14.3t	97.50% <97.50%> (+<0.01%)	⬆️
Py-pypy3.11.15-7.3.21	97.40% <22.50%> (-0.14%)	⬇️
VM-macos	97.87% <98.75%> (+<0.01%)	⬆️
VM-ubuntu	98.71% <98.75%> (ø)
VM-windows	96.98% <98.75%> (+<0.01%)	⬆️

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.

[codspeed-hq]

Merging this PR will not alter performance

✅ 59 untouched benchmarks

---

_{Comparing josumoreno-BP:fix/zstd-multiframe (9a000fd) with master (07a2ca4)}

[Dreamsorcerer]

I think we're missing an edge case, which probably needs a new test.

If we have unused data here and break, we've ended up throwing it away. Presumably when it continues with further data, it'll fail as the input is corrupted.

[Dreamsorcerer]

Other than that subtle edge case, this looks very solid!