arb-2.21.1-bp154.1.15

- Update to release 2.21.0
  * Experimental new arb_fpwrap module: accurate floating-point
    wrappers of Arb mathematical functions (supersedes the
    external arbcmath.h).
  * Fixed memory leak in arf_load_file.
  * New and faster gamma function code.
  * Most gamma function internals are now located in the
    arb_hypgeom and acb_hypgeom modules. The user-facing
    functions (arb_gamma, etc.) are still available under the old
    names for compatibility. The internal algorithms for rising
    factorials (binary splitting, etc.) have been moved without
    aliases.
  * Added arb_fma, arb_fma_arf, arb_fma_ui (like addmul, but take
    a separate input and output).
  * Slightly faster internal Bernoulli number generation for small n.
  * Better enclosures for acb_barnes_g at negative reals.
  * Added Graeffe transforms (arb_poly_graeffe_transform,
    acb_poly_graeffe_transform).
  * Added acb_add_error_arb.

- Update to release 2.19.0
  * Significant improvements to the implementation of Platt's
    algorithm for computing Riemann zeta function zeros at large
    height.
  * Better criterion for selecting asymptotic expansion of
    incomplete gamma function (contributed by p15-git-acc).
  * Multithreaded acb_dft for power-of-two lengths.
  * Added acb_csc_pi, arb_csc_pi.
  * Fixed segfault in acb_mat_eig_simple_rump when called with L
    non-NULL and R NULL.
  * Fixed bug in acb_real_abs.
  * Changed several functions to more consistently return
    infinities instead of NaNs where reasonable.
  * Added Fransen-Robinson as an integral example.

- Update to release 2.18.1
  * Faster and more accurate real matrix multiplication using
    block decomposition, scaling, and multiplying via FLINT
    integer matrices in combination with safe use of doubles for
    radius matrix multiplications.
  * New C functions for Riemann zeta function.
  * For more, see the history.rst file in the arb-devel subpackage.

- Update to new upstream release 2.13.0
  * Fixed rounding direction in arb_get_abs_lbound_arf() which in
    some cases could result in an invalid lower bound being
    returned.

- Update to new upstream release 2.12.0

- Update to new upstream snapshot 1.0.0+git7

- Initial package (version 0.5+git28) for build.opensuse.org