初版：实现 AMESim 机械因果化与事件求解

初步支持 MECMAS21 刚性质量状态归并、端止事件、恢复系数，以及 LSTP 接触和压力流量显式因果化。

已知问题：显式传播仍会重复扫描全网方程，长时刚性仿真性能待优化；自适应积分器遇到越出物理域的试探状态时，尚未实现恢复并缩步重试。

tests/test_pressure_flow_solver_initialization.py

@@ -127,7 +127,7 @@ class PressureFlowSolverInitializationTests(unittest.TestCase):
 
         with patch.object(
             PressureFlowSolver,
-            "_seed_explicit_mass_flows",
+            "_solve_explicit_flow_unknowns",
             return_value=None,
         ), patch("scipy.optimize.least_squares", side_effect=exact_status_zero):
             diagnostics = exact_solver.solve()
@@ -148,7 +148,7 @@ class PressureFlowSolverInitializationTests(unittest.TestCase):
 
         with patch.object(
             PressureFlowSolver,
-            "_seed_explicit_mass_flows",
+            "_solve_explicit_flow_unknowns",
             return_value=None,
         ), patch("scipy.optimize.least_squares", side_effect=inaccurate_status_zero):
             with self.assertRaises(AlgebraicSolveError):
@@ -168,7 +168,7 @@ class PressureFlowSolverInitializationTests(unittest.TestCase):
 
         with patch.object(
             PressureFlowSolver,
-            "_seed_explicit_mass_flows",
+            "_solve_explicit_flow_unknowns",
             return_value=None,
         ), patch("scipy.optimize.least_squares", side_effect=exact_invalid_status):
             with self.assertRaises(AlgebraicSolveError):