Fixed accuracy in some cases

[sampad1370]

Also, it's improved with removing dividing and multiplying with "total" in every step.

It will be fixed incorrect result in the following example:

auto tween = tweeny::from(1.0f, 1.0f)
.to(7.0f, 7.0f)
.during(60)
.via(tweeny::easing::enumerated::linear)
.onStep([](float i, float f) {
printf("running: i=%f f=%f\n", i, f);
fflush(stdout);
return false;
});
while (tween.progress() < 1.0f) {
auto [value1, value2] = tween.peek(tween.progress());
printf(" progress:%f and values is: %f,%f\n", tween.progress(), value1, value2);
fflush(stdout);
tween.step(10);
}
fflush(stdout);
tween.backward();
while (tween.progress() > 0.0f) {
auto [value1, value2] = tween.peek(tween.progress());
printf(" progress:%f and values is: %f,%f\n", tween.progress(), value1, value2);
fflush(stdout);
tween.step(10);
}
auto [value1, value2] = tween.peek(tween.progress());
printf(" progress:%f and values is: %f,%f\n", tween.progress(), value1, value2);

I get following result on Ubuntu 22.04 with gcc-11 without this commit:

progress:0.000000 and values is: 1.000000,1.000000
running: i=2.000000 f=2.000000
progress:0.166667 and values is: 2.000000,2.000000
running: i=3.000000 f=3.000000
progress:0.333333 and values is: 3.000000,3.000000
running: i=4.000000 f=4.000000
progress:0.500000 and values is: 4.000000,4.000000
running: i=5.000000 f=5.000000
progress:0.666667 and values is: 5.000000,5.000000
running: i=6.000000 f=6.000000
progress:0.833333 and values is: 6.000000,6.000000
running: i=7.000000 f=7.000000
progress:1.000000 and values is: 7.000000,7.000000
running: i=6.000000 f=6.000000
progress:0.833333 and values is: 6.000000,6.000000
running: i=4.900000 f=4.900000
progress:0.666667 and values is: 4.900000,4.900000
running: i=3.900000 f=3.900000
progress:0.500000 and values is: 3.900000,3.900000
running: i=2.900000 f=2.900000
progress:0.333333 and values is: 2.900000,2.900000
running: i=1.900000 f=1.900000
progress:0.166667 and values is: 1.900000,1.900000
running: i=1.000000 f=1.000000
progress:0.000000 and values is: 1.000000,1.000000

but it must be:

progress:0.000000 and values is: 1.000000,1.000000
running: i=2.000000 f=2.000000
progress:0.166667 and values is: 2.000000,2.000000
running: i=3.000000 f=3.000000
progress:0.333333 and values is: 3.000000,3.000000
running: i=4.000000 f=4.000000
progress:0.500000 and values is: 4.000000,4.000000
running: i=5.000000 f=5.000000
progress:0.666667 and values is: 5.000000,5.000000
running: i=6.000000 f=6.000000
progress:0.833333 and values is: 6.000000,6.000000
running: i=7.000000 f=7.000000
progress:1.000000 and values is: 7.000000,7.000000
running: i=6.000000 f=6.000000
progress:0.833333 and values is: 6.000000,6.000000
running: i=5.000000 f=5.000000
progress:0.666667 and values is: 5.000000,5.000000
running: i=4.000000 f=4.000000
progress:0.500000 and values is: 4.000000,4.000000
running: i=3.000000 f=3.000000
progress:0.333333 and values is: 3.000000,3.000000
running: i=2.000000 f=2.000000
progress:0.166667 and values is: 2.000000,2.000000
running: i=1.000000 f=1.000000
progress:0.000000 and values is: 1.000000,1.000000

The result of direct stepping and backward stepping is different! And I think there are many similar cases. And I think extra divining and multiplying can be removed.
Please check my commit.

[mobius3]

Thanks! This is very nice and I'm willing to accept it, but there are some unrelated edits, mostly in regards to formatting, that I don't think are justifiable for what this PR does (it's even harder to review because of that). I was going to point it out directly at each inconsistency but I suggest, at least for this PR, turning off code formatting in your IDE, reverting the code to what it was previously and performing your changes again, without breaking code style.