// TODO: still to do, possibly: median, percentile, corr, cov

// Sample covariance of two series (aligned on keys, inner join)
Stats.cov s1 s2 : float

// Pearson correlation coefficient of two series
Stats.corr s1 s2 : float

let prices = series [ "MSFT" => 300.0; "AAPL" => 180.0; "GOOG" => 140.0 ]
let returns = series [ "MSFT" =>  0.05; "AAPL" =>  0.03; "GOOG" =>  0.04 ]

Stats.cov prices returns    // sample covariance
Stats.corr prices returns   // Pearson r
```

## Design

The implementation follows the same conventions as all other `Stats` members: `inline`, single-pass via `Array.ofSeq`, and operates on the `Series.ZipInner` result so that missing-value semantics are handled consistently with the rest of the library.

## Test Status

Build and all 515 tests pass (508 existing + 7 new, covering correct values, NaN edge cases, key alignment, perfect/anti-correlation, and zero-variance handling).

```
Passed! — Failed: 0, Passed: 515, Skipped: 0, Total: 515