Add overflow protection and input validation for mathematical operations

[Copilot]

Reviewed codebase for mathematical operator errors that could cause runtime exceptions. While the specific landslide_threshold function mentioned in the issue no longer exists, identified and protected vulnerable operations in the current implementation.

Changes

• Overflow/underflow guards in landslide_probability()

  • math.exp() overflows at z≈710; added guards at ±700 threshold
  • Early return 1.0 for z>700, 0.0 for z<-700
  • Optimized from two exp() calls to one

• Input validation in lambda_handler()

  • Clamp rainfall to [0, 2000]mm range
  • Prevents negative values (sensor errors)
  • Prevents overflow in downstream calculations (24hr/2day/3day projections multiply by 8/16/24)

• Named constants

  • Z_OVERFLOW_THRESHOLD = 700
  • MAX_VALID_RAINFALL_MM = 2000

# Before
return math.exp(z) / (1 + math.exp(z))  # Two exp() calls, can overflow

# After  
if z > Z_OVERFLOW_THRESHOLD:
    return 1.0
elif z < -Z_OVERFLOW_THRESHOLD:
    return 0.0
exp_z = math.exp(z)
return exp_z / (1 + exp_z)  # One exp() call, overflow protected

Note: Silent clamping may mask data quality issues. Consider adding logging in production to detect sensor malfunctions.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.