add some speed-up tricks

[SkySailing]

电脑硬件：
Intel(R) Xeon(R) CPU E5-2678 v3 @ 2.50GHz

final result:
6285 ms / 1096 ms = 5.734

优化过程如下：
原始版本
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 6285 ms

-o1 优化
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2659 ms

-o2
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2558 ms

-o3
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2560 ms

索引int变成size_t
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2560 ms

inline frand函数
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2593 ms

将/2改成*0.5
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2585 ms

将公共除法变成乘法
Initial energy: -13.414000
Final energy: -13.356842
Time elapsed: 2030 ms

-ffast-math 选项定义了预处理器宏 FAST_MATH, 指示编译不必遵循 IEEE 和 ISO 的浮点运算标准
-march=native，GCC会自动检测你的CPU支持的指令集。
Initial energy: -13.414000
Final energy: -13.356841
Time elapsed: 1606 ms

SoA 阵列结构 struct of array
Initial energy: -13.414000
Final energy: -13.356841
Time elapsed: 1630 ms

sqrt前+std:: 使用模板完成传入参数匹配
Initial energy: -13.414000
Final energy: -13.356841
Time elapsed: 1220 ms

#pragma GCC unroll 4
Initial energy: -13.414000
Final energy: -13.356841
Time elapsed: 1212 ms

G * dt / d2 提取到循环外面计算
Initial energy: -13.414000
Final energy: -13.356841
Time elapsed: 1096 ms