[Repo Assist] feat: expose QR decomposition with GramSchmidt and Householder method choice

[github-actions]

🤖 This is an automated PR from Repo Assist.

Closes #317

Problem

FSharp.Stats had no user-facing API for choosing the QR decomposition method. Users who wanted the Gram-Schmidt variant had to reach down into FsMath.Algebra.LinearAlgebra.qrModifiedGramSchmidt directly, bypassing any FSharp.Stats documentation or discoverability.

Solution

New FSharp.Stats.LinearAlgebra module with:

Symbol	Description
QRMethod.Householder	Full QR via Householder reflections; Q is m×m, R is m×n
QRMethod.GramSchmidt	Thin/economy QR via modified Gram-Schmidt; Q is m×n, R is n×n
QR.decompose method A	Method-dispatching entry point
QR.householder A	Named shorthand for Householder
QR.gramSchmidt A	Named shorthand for Gram-Schmidt

Example

open FsMath
open FSharp.Stats.LinearAlgebra

let A = matrix [[12.;-51.;4.];[6.;167.;-68.];[-4.;24.;-41.]]

// Gram-Schmidt – thin Q (m×n)
let (qGS, rGS) = QR.decompose GramSchmidt A   // Q: 3×3, R: 3×3 (square input)

// For a 4×3 input the difference is visible:
let B = matrix [[1.;2.];[3.;4.];[5.;6.];[7.;8.]]
let (qThin, rThin) = QR.decompose GramSchmidt B   // Q: 4×2, R: 2×2
let (qFull, rFull) = QR.decompose Householder B   // Q: 4×4, R: 4×2
```

## Trade-offs

- No new algorithms introduced; both methods delegate to `FsMath.Algebra.LinearAlgebra` which is already a dependency.
- Gram-Schmidt (thin Q) uses less memory than Householder (full Q) for tall matrices, which is a common use case in regression.
- Both variants satisfy A = Q * R and Q^T Q = I; the only difference is the shape of Q and R.

## Test Status

10 new tests added:

- `A = Q * R` round-trips (both methods, square and rectangular matrix)
- `Q^T Q = I` orthonormality (both methods)
- R upper-triangular (both methods)
- Dimension checks: GramSchmidt gives thin Q (m×n), Householder gives full Q (m×m)
- `QR.decompose` dispatch test

```
Passed!  - Failed: 0, Passed: 1203, Skipped: 0, Total: 1203

Generated by 🌈 Repo Assist, see workflow run. Learn more.

To install this agentic workflow, run

gh aw add githubnext/agentics/workflows/repo-assist.md@7ee2b60744abf71b985bead4599640f165edcd93