fix: return AbortError instead of TypeError when consuming aborted cloned response

[mcollina]

When a Response was cloned and the AbortController was aborted before consuming the body, consuming the body threw a TypeError ('Body is unusable') instead of the expected AbortError.

Changes

• Check for aborted state before checking if body is unusable in consumeBody()
• Set response.aborted flag in abortFetch() to properly track abort state

Test

Includes test from #4800 (thanks @afoures)

Fixes: #4799
Refs: #4800

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.25%. Comparing base (352c691) to head (f678a03).
⚠️ Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4801      +/-   ##
==========================================
- Coverage   93.25%   93.25%   -0.01%     
==========================================
  Files         109      109              
  Lines       34024    34105      +81     
==========================================
+ Hits        31729    31803      +74     
- Misses       2295     2302       +7     

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[KhafraDev]

none of the code here is correct

[KhafraDev]

What I've noticed from using Claude on the code in undici: it sucks by default. I've found ways to improve on its output while also speeding up my process for bug fixing; have it fetch the spec in its entirety and then ask it to 1) do not edit existing comments under any circumstance, and 2) implement changes verbatim to the spec text. Without the spec text, it often does things like we see here where it adds a non-existent step 6 with a nonsensical change.

I usually end up doing all of the coding manually, using Claude to track down bugs, but then implementing the fix myself to ensure Claude doesn't muck it up.