Release v0.6.5

[lkdvos]

MatrixAlgebraKit.jl v0.6.5

This release completes the separation of Algorithm and Driver concepts across all decomposition types, and adds several new features including SafeDivideAndConquer, TruncationUnion, DefaultAlgorithm, and AD rules for hermitian projections.

Highlights

• Algorithm/Driver separation for QR/LQ, SVD, and Schur/Eig decompositions — cleaner dispatch and easier extensibility (Separate Algorithm and Driver #178, Separate Algorithm and Driver - part II (SVD) #189, Separate Algorithm and Driver - part III (Schur-Eig) #194, some more updates on algorithms #196)
• SafeDivideAndConquer — a more robust SVD algorithm variant (SafeSVD alternative #185)
• TruncationUnion — combine multiple truncation schemes with minimal rank support (Add TruncationUnion to support supplying minimal ranks #183)
• DefaultAlgorithm — explicit dispatch to the default algorithm (Add DefaultAlgorithm #195)
• AD rules for (anti-)hermitian projection via ChainRules, Mooncake, and Enzyme (AD rules for (anti-) hermitian projection #174)

Full Changelog

See CHANGELOG for the complete list of changes.

[lkdvos]

Before merging, let me locally check that TensorKit/MPSKit still work, i.e. that all deprecations etc are handled correctly.

[edit] seems to work for me, I am OK to merge and release this!