Fix promoter/attI coordinates on linear contigs

[djken2009]

This PR fixes the coordinate overflow issue for P_intI promoter and attI site predictions on linear contigs, as reported in issue #120.

Problem

• Kpne_VR_40131: P_intI1 predicted at 188068-188102, but contig_3 only has 187829 bp
• Ecol_VR_96159: P_intI1 predicted at 78180-78214, but contig_8 only has 77917 bp

Solution

• Added topology-aware boundary checking using max() and min() functions
• Separate handling for circular vs linear topologies
• Added comprehensive test cases

Testing

• ✅ All 248 tests pass
• ✅ Coordinates now within valid ranges

Closes #120