Optimize reference genome loading memory usage by using string instea…

[cascadingstyletrees]

💡 What: Changed CACHED_REFERENCE_FILE in yleaf/Yleaf.py from a List[str] (where each element was a character) to a single str. Updated the load_reference_file function to construct this string efficiently using join.

🎯 Why: Storing a large genome sequence as a list of individual characters incurs massive memory overhead in Python. Using a single string is far more efficient.

📊 Measured Improvement:
Benchmark with a 10MB generated FASTA file:

• Memory Usage: Reduced from ~89.45 MB to ~26.21 MB (~70% reduction).
• Execution Time: Reduced from ~1.03s to ~0.67s (~35% improvement).
• Correctness: Verified that the result type is str and indexing/content matches expectations.