feat(metric): Add CUFD (Codon Usage Frequency Distribution) metric and comprehensive test suite

[Rakshitha-Ireddi]

This PR introduces the Codon Usage Frequency Distribution (CUFD) metric to the ARCADE calculator module, along with a comprehensive test suite covering all calculator functions.

Key Changes:

• New Metric: Added cufd_kl_divergence and cufd_cosine_similarity to calculator/calculator.py. These metrics measure how closely a sequence matches a target organism's full codon usage distribution.
• Data: Added calculator/data/codon_usage_human.json containing human codon frequencies from the Kazusa database.
• Steering: Extended scripts/utils/mutation.py with a cufd mode to generate contrastive pairs (high vs. low CUFD) for steering vector computation.
• Testing: Added a full test suite in tests/ (previously nonexistent) covering:
  All calculator metrics (MFE, CAI, GC, CpG, UpA, CUFD)
  Codon<->Amino Acid mappings
  Tokenizer utilities

Verification

Added tests/ directory with 61 tests.
Run pytest tests/ to verify. All tests pass locally.

Authors

• IreddI Rakshitha
• Yaswanth Devavarapu