Add diagnostics ; evaluate row/column p-value separately

[orionlee]

They are two loosely related issues.

1. Support diagnostics by providing ways to access the estimated centroids, smoothed centroids, etc. with new include_diagnostics parameter in centroid_test() function.

2. The p-value currently used for centroid shift detection is a mean of p-values of the X and Y centroids. That seems to be problematic.

E.g., for TIC 13023738, sector 2 referenced in the associated paper, centroid_test() reports no centroid shift detected because the mean p-value is large (~0.25). But in reality, there is clear centroid shift in X (tiny p-value ~6e-7). In the test, it is overshadowed / averaged out by the large p-value in Y (~0.50).

To handle such cases, the second commit of this PR changes the detection from using a mean of X / Y p-values, to using the minimum of X and Y p-values.

---

E.g., for TIC 13023738, sector 2 referenced in the associated paper, the diagnostics could help to pinpoint the X / column centroids have a noticeable shift:

Another use is to manually inspect whether the smoothing is over/under aggressive.

---

Note: If the PR is to be accepted, some polish probably needs to be done (e.g., documentation, consistent labeling of the planet candidate, deciding what p-value(s) to report, etc.). I'm holding off pending feedback.