[tatikoma]

Summary

mod_cache_disk serializes each cache metadata entry as one logical line. On recall, read_table() and read_array() used apr_file_gets() with a fixed buffer and parsed every returned chunk as a complete entry. A line longer than the buffer could therefore be split into multiple entries.

This change accumulates chunks through the terminating LF before parsing the line, for both response-header tables and Vary arrays. The regression primes a disk-cache entry with a response header spanning the old read boundary and verifies that recall preserves the complete value as one field.

Testing

• Unpatched current trunk: existing cache test passed; new regression failed.
• Patched current trunk: test/pytest_suite/tests/t/modules/test_cache.py — 2 passed.

---

• This PR does not report or address a security vulnerability
  (see SECURITY.md and
  https://www.apache.org/security/#reporting-a-vulnerability).
• Code follows the httpd style guide.
• New log messages (level debug or higher) use an empty APLOGNO() tag;
  numbers are assigned by a committer at merge time (see docs/log-message-tags/README).
• If the change is user-visible, a changes-entries/*.txt file is included,
  following the template in README.CHANGES (not needed otherwise).
• Test cases are included where appropriate.