[BUG] Harden `malloc_aligned_4k` (Overflow Guard + Quiet Failure)

[k8ika0s]

Proposed PR branch: qe/issue-06-malloc4k-hardening

Category: Bug (memory safety hardening)

Problem

• malloc_aligned_4k() must reject sizes that overflow internal calculations.
• Library code paths should not print to stderr on allocation failures.

Justification / why this is needed

• Overflow in allocation-size math is a classic memory safety footgun; even if
  not exploitable, it can cause crashes and silent data corruption.
• Printing to stderr from a library is hostile to integrators (pollutes logs,
  breaks structured logging, and can leak details in restricted environments).
• Baseline+regtests showed the overflow guard + quiet-failure expectations
  failing, while the enhanced tree passes.

Proposed change

• Add explicit overflow checks.
• Remove stderr prints on failure paths.
• Add regression tests for overflow guard and “quiet failure”.

Acceptance criteria

• Overflow-size requests return NULL.
• No unexpected stderr output for allocation failure paths.
• Regression tests cover both behaviors.

Test plan

• make test (expects tests/testDriver_malloc4k.c to pass).

References

• Baseline report: artifacts/FINAL_BASELINE_ENHANCEMENT_REPORT.md shows two
  malloc4k regressions failing on baseline+regtests and passing in enhanced.
• Key files: zdnn/malloc4k.c, tests/testDriver_malloc4k.c.