Release 1.5

[cmutel]

Summary

• Add ArrayEntry dataclass and add_array_entries method (PR Add ArrayEntry dataclass and add_array_entries method #102)
• Prune trivial distributions arrays in finalize_serialization (PR Prune redundant distributions arrays in finalize_serialization #101)
• Add rescale to MatrixEntry/ArrayEntry and rename scale_array → rescale_array (PR Add rescale to MatrixEntry/ArrayEntry and rename scale_array → rescale_array #104)
• Fix data-dependent flip_array dtype check in all add_* methods (PR Fix data-dependent flip_array dtype check in all add_* methods #103)
• Deduplicate chunked bucket-fill logic; fix Bug: create_chunked_array uses np.hstack instead of np.vstack, producing wrong shape for >500-row inputs #95 and Efficiency: partial-chunk slice in create_chunked_* is a view, keeping full bucket buffer alive until final concatenation #97 (PR Deduplicate chunked bucket-fill logic; fix #95 and #97 in the process #100)

Test plan

• Run full test suite: pytest
• Verify version is 1.5 in bw_processing.__version__
• Check changelog entry is correct in CHANGES.md