- Update to 1.11.4:
* MAINT: ensure cobyla objective returns scalar
* MAINT: fixup dep warning
* BUG: interpolate: fix spalde with len(c) < len(t)
* BUG: pass unused xrtol in fmin_bfgs to _minimize_bfgs
* BUG: Regression test for lsq trf boundary error
* BUG: lsq trf gives x=1e-10 if x0 is near a bound
* BUG: make Bessel-roots function not hang and not skip roots
* io/matlab: Fix loading of mat files containing fn handles when
simplify_cells=True
* BUG: make L-BFGS-B work with single precision gradient
* MAINT: stats: fix NumPy DeprecationWarnings
* BUG: sparse.linalg: Cast to intc before SuperLU
* MAINT: Use deb_system scheme to match meson's path inference & fix aarch64
* BUG: Fix python3.12 distutils dev.py build
* BUG: stats: remove use of `BOOST_MATH_DECLARE_SPECIAL_FUNCTIONS`
* MAINT: fix libquadmath licence
* MAINT: newton, make sure x0 is an inexact type
* MAINT: stats.CovViaEigendecomposition: fix _colorize for singular
covariance
* TST: fix `TestODR.test_implicit` test failure with tolerance bump
* BUG: signal: freqz rfft grid fix
* BUG: Support sparse arrays in scipy.sparse.csgraph.laplacian
* MAINT: signal: Remove the cval parameter from the private function
_pad_test.
* BLD: Avoid absolute pathnames in .pyx files
* BUG: Add back make_strictly_feasible to lsq trf.
* MAINT: should not be using np.float64() on arrays
* BUG: trust-constr Bounds exclusive
* BUG: sparse.csgraph: Support int64 indices in traversal.pyx