v0.1.4: fix :Out-of-bounds read in simd_search (#2)

Fixed

• Prevent buffer overrun in simd_search:
  • Added a check to ensure at least 16 bytes remain before performing _mm_loadu_si128.
  • Implemented a scalar fallback for handling remaining bytes safely.
  • Fixes #2.

Improved

• Optimized SIMD search logic:

  • Reordered conditions to enhance branch prediction.
  • Ensured correct handling of overlapping matches.

• Memory safety improvements:

  • Replaced direct modification of pattern with strdup().
  • Ensured allocated memory (clean_pattern) is properly freed after use.