Major updates of the arithmetic PCV code

[HanatoK]

The new code has following improvements with respect to the old code:

1. Correctly calculate s and z with logsumexp and shifted softmax to
   avoid overflow and underflow;
2. Correctly calculate the derivatives of s and z with respect to vector
   sub CVs (also use logsumexp and shifted softmax);
3. Implement debug gradients for s and z with respect to sub CVs that
   have no explicit gradients;
4. Add new tests for derivatives;

The new aspath and azpath also implement the PCV calculation using Parrinello's formula in atomic Cartesian space.

[HanatoK]

This implementation compute $s$ using $\exp(\mathrm{LSE}_0 - \mathrm{LSE}_1)$, which seems OK in all my tests and my simulation crashed with direct calculations. An alternative way could be summing up the shifted-softmax. According to this paper, in case of numerical stability, using

$$ g_j = \frac{{{e}}^{x_j-a}}{ \sum_{i=1}^n{{e}}^{x_i-a}} $$

is better than

$$ g_j = \exp\left(x_j - \log\sum_{i=1}^n{{e}}^{x_i}\right) $$

I am not sure if it is better in the case of calculating $s$.

[jhenin]

I think our main concern here is stability and performance. Since we use double precision, accuracy should be ok once we avoid under/overflow.

[HanatoK]

I think our main concern here is stability and performance. Since we use double precision, accuracy should be ok once we avoid under/overflow.

Thanks @jhenin ! The sum of shifted-softmax is used in this PR. As for performance I think the two ways are basically the same.

[giacomofiorin]

Great changes.

It would be better if the tests were included in the top folder rather than being specific to the NAMD sub-folder. However, this would require some more effort to handle the additional files.