Fix double bounds in MinMax causing unsatisfiable constraints and false SAFE results

[shawshank-202]

Fixes #34
This PR corrects the bounds for symbolic double variables in MinMax, which resulted in an invalid range and unsound constraint generation.

---

Problem

Symbolic double bounds were defined as:

    symbolic.min_double = Double.MIN_VALUE
    symbolic.max_double = Double.MAX_VALUE

However, Double.MIN_VALUE is the smallest positive value (~4.9e-324), not the most negative value.
This excludes all negative values from the domain.

---

float-nonlinear-calculation/Optimization2, JPF generates constraints like:

    (assert (>= double0 (/ 1.0 0.0)))
    (assert (<= double0 0.0))

This leads to:

    double0 ≥ +∞ && double0 ≤ 0  → UNSAT

All explored path conditions become unsatisfiable:

    ### PCs: total:3 sat:0 unsat:3
    ***********Warning: everything false

---

Result Before Fix

SAFE

No paths are explored, leading to an incorrect SAFE verdict.

---

Results can be verified against
float-nonlinear-calculation/Optimization2
float-nonlinear-calculation/Optimization5
float-nonlinear-calculation/Optimization6
float-nonlinear-calculation/coral81
float-nonlinear-calculation/coral84

[shawshank-202]

Hey @sohah @vaibhavbsharma can you have a look at this and kindly give your feedback, thank you.