Update Eigen to commit:cd4f989f8f9288ab5aed1643ecb04c7be021021e

CHANGELOG
=========
cd4f989f8 - assume_aligned uses bytes not bits
ac7c192e1 - Add a bunch of useful scripts for planning releases.
5bc944a3e - Fix jacobi svd for TriangularBase
dbe9e6961 - Fix BLAS/LAPACK DLL usage on Windows.
ef3c5c1d1 - Add workaround for using std::fma for scalar multiply-add.
5996176b8 - Fix alignment bug in avx pcast<Packet4l, Packet4d>
4bd382df5 - Fix SSE PacketMath Compilation Error on QNX
13bd14974 - fix errors in windows builds and tests
eea6587b0 - Fix scalar_inner_product_op when binary ops return a different type
7eaf9ae68 - Add a method to SelfAdjointEigenSolver for computing the matrix exponential
32b0f386b - Eliminate possible -Wstringop-overflow warning in .setZero()
1edf360e3 - Fix line endings
ccde35bcd - Update dev version number.
a67f9dabb - tests: add missing link
e6792039f - Update changelog to reflect 3.4.1 and 5.0.0 releases.
4916887f2 - Update `geo_homogeneous` test, add eval() to PermutationMatrix.
5c1029be1 - The 'CompressedStorageIterator<>' needs to satisfy the RandomAccessIterator
f9f515fb5 - get rid of a bunch of windows jobs
2e5447e62 - STYLE: Scripts with shebang should be executable
8d7ebac6e - Disambiguate multiplication of a permutation matrix and a homogeneous vector
bea7f7c58 - SparseMatrixBase: delete redundant/shadowed typedef
7292c78e1 - blas: Fix parenthesis suggestion warning
e524488eb - Convert Mercurial hgeol to gitattributes
dbd25f632 - Fix select: return typed comparisons if vectorized
027dc5bc8 - Extend the range of supported CMake package config versions
4df215785 - Support matrix multiplication of homogeneous row vectors
2d170aea1 - Define pcmp_le generically in terms of pcmp_eq and pcmp_lt.
ea869e183 - Add missing bool SSE2 PacketMath comparison
6ef18340a - CMake: Explicit STATIC libs
14477c5d4 - Replace deprecated std::is_trivial by an internal definition
b2ec79a23 - Move smoketests to small GitLab runners.
62fbd276e - Provide hints for deprecated functionality
d38d669fd - Fix real schur exceptional shift
4ac3e71f7 - CMake: Require at least C++14
a627f72cd - Add "Version" file and update version.
e0a59e5a6 - Fix typo
6cd6284f7 - Make the sparse matrix printer pretty
e5f3fa2d6 - Add gemmtr implementation.
f426eff94 - Add inline/device-function attributes to fma.
da1a34a6b - Zero-out matrix for empty set of triplets.
52fc978c6 - fixed typo `sparcity` -> `sparsity`
8a8fbc8f5 - Don't enable AVX for wasm.
70d8d99d0 - Only build docs on push to main branch, not MRs.
7f0cb638c - Specialize numext::madd for half/bfloat16.
1e9d7ed7d - Add missing semicolon to `has_fma` definitions to fix GPU builds.
5d4485e76 - Move more jobs to gitlab runners.
2e8cc042a - Replace calls to numext::fma with numext:madd.
52f570a40 - Move GPU ci jobs to gitlab-hosted runners.
38b51d5b7 - Mitigate setConstant regression with custom scalars
d2a70fe4e - Make permutation products aliasing by default.
4ae564735 - Fix direct index aliased assignment.
1a45d2168 - Fix use of FMA in triangular solver for boost multiprecision.
05e74b1a4 - Tiny fix in `unsupported/Eigen/CXX11/src/Tensor/TensorContraction.h`
d36899812 - Fix MSVC error about missing std::bit_cast
c487a4fe9 - Clean up most of testsuite on s390x

PiperOrigin-RevId: 820448272
Change-Id: Ie30f90f50d513e695d90b1295061104a87e8a762