refactor cvector_at

[GoodenoughPhysicsLab]

After I closed my issue, I suddenly had an idea.

Now, cvector_at can be used as follow:

#include "cvector.h"
#include <stdio.h>
#include <assert.h>

int main(void) {
    cvector(int) v = NULL;
    cvector_push_back(v, 1);

    printf("%d\n", cvector_at(v, 0)); // get value
    cvector_at(v, 0) = 2; // set value
    printf("%d\n", cvector_at(v, 0));
    // printf("%d\n", cvector_at(v, 1)); // assert fail

    return 0;
}

[eteran]

Hmm, interesting usage of assert and comma operator there. I'll take a closer look in a little while

[GoodenoughPhysicsLab]

I have fixed all the test(including some bug of test e.g. calloc(sizeof(struct data_t), 1)

[eteran]

Can you explain what you believe was broken about:

struct data_t *data = calloc(sizeof(struct data_t), 1);

vs what you changed it to:

struct data_t *data = calloc(1, sizeof(struct data_t));

Given that the nmem and size parameters are basically just multiplied (and checked for overflow), this should make no difference. I don't really mind the change, I was just curious about the reasoning.

[eteran]

Also, it appears that your PR requires adding an #include to the cvector.h file.

[GoodenoughPhysicsLab]

Can you explain what you believe was broken about:

The compiler gives me warning.

void *__cdecl calloc(size_t _NumOfElements,size_t _SizeOfElements);

[eteran]

Thanks!