Submit size and depth record candidates for voter

[aandriasyan]

This PR submits two candidate implementations for the voter benchmark.

Files:

• best_results/size/voter_size_2026.blif

  • claimed LUT count: 993
  • claimed depth: 12

• best_results/depth/voter_depth_2026.blif

  • claimed LUT count: 1000
  • claimed depth: 10

Both files follow the benchmark BLIF convention:

• .model top
• inputs A[0]..A[1000]
• output maj
• internal nodes named n...

Local verification:

• ABC reads both submitted BLIF files without interface/naming errors.
• ABC reports:
  • voter_size_2026.blif: 993 nodes, 12 levels
  • voter_depth_2026.blif: 1000 nodes, 10 levels
• ABC miter simulation against random_control/voter.blif passed with 100000 frames and 64 words.
• Exact-count randomized verification passed: 1000 randomly shuffled vectors for every Hamming weight 0..1001 for each submitted file.

Method:
SymFold is a constructive LUT-K synthesis method for symmetric Boolean functions.
For this submission, the method is instantiated for K=6 and applied to the
voter benchmark, which is a threshold/majority function over 1001 inputs.

Submitted by Aleksandr Andriasyan, independent researcher.