added cholesky of cholesky (#55559)

Currently, if you have a cholesky matrix, you cannot call `cholesky` on
it again:

```
using LinearAlgebra

N = 10
A = randn(N, N)
P = Symmetric(A * A' + I)

C = cholesky(P)

CC = cholesky(C)   # this line throws an error
```

This small PR provides the fix.

---------

Co-authored-by: Jeff Bezanson <jeff.bezanson@gmail.com>

Author: dev10110

stdlib/LinearAlgebra/src/cholesky.jl

@@ -551,6 +551,9 @@ end
 # allow packages like SparseArrays.jl to hook into here and redirect to out-of-place `cholesky`
 _cholesky(A::AbstractMatrix, args...; kwargs...) = cholesky!(A, args...; kwargs...)
 
+# allow cholesky of cholesky
+cholesky(A::Cholesky) = A
+
 ## With pivoting
 """
     cholesky(A, RowMaximum(); tol = 0.0, check = true) -> CholeskyPivoted

stdlib/LinearAlgebra/test/cholesky.jl

@@ -630,4 +630,14 @@ end
     end
 end
 
+@testset "cholesky_of_cholesky" begin
+    for T in (Float64, ComplexF64), uplo in (:U, :L)
+        A = randn(T, 100, 100)
+        P = Hermitian(A' * A, uplo)
+        C = cholesky(P)
+        CC = cholesky(C)
+        @test C == CC
+    end
+end
+
 end # module TestCholesky