Fix decode_array double-reading length bytes for arrays with 24+ items

[cffls]

The custom decode_array override in serialization.py called _decode_length to check for indefinite-length arrays, then delegated to the original decode_array which called _decode_length again. For arrays with fewer than 24 items, the length is encoded directly in the subtype (no stream bytes consumed), so the double call was harmless. For 24+ items, CBOR uses multi-byte length encoding (e.g. 98 18 for 24 items) and _decode_length reads from the stream — the second call consumed actual array content as a length byte, corrupting the decode.

Replace the _decode_length call with a simple subtype == 31 check, which is sufficient to detect indefinite-length arrays without consuming any bytes from the stream.

This bug only affected cbor2pure, not the cbor2 C extension.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.60%. Comparing base (8946e19) to head (d342861).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #479      +/-   ##
==========================================
- Coverage   90.61%   90.60%   -0.01%     
==========================================
  Files          34       34              
  Lines        5155     5154       -1     
  Branches      781      781              
==========================================
- Hits         4671     4670       -1     
  Misses        305      305              
  Partials      179      179              

☔ 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.