chore: release v0.2.16

[github-actions]

🤖 New release

• libm: 0.2.15 -> 0.2.16 (✓ API compatible changes)

Changelog

0.2.16 - 2025-12-07

Other

• Update Cargo.toml authors fields
• Implement exp and its variants for i586 with inline assembly
• Optimize fmod
• Document the purpose of a helper module
• Implement accelerated computation of (x << e) % y in unsigned integers
• require some basic operators in trait DInt
• Compare against CARGO_CFG_TARGET_FAMILY in a multi-valued fashion
• Add back the unsafe for intrinsics::fma but allow(unused_unsafe)
• Merge ref 'caccb4d0368b' from rust-lang/rust
• Remove usage of the to-be-deprecated core::f32, core::f64 items
• define and implement trait NarrowingDiv for unsigned integer division
• Fix some typos
• Use CARGO_CFG_*_{F16,F128} rather than invoking rustc
• Simplify the configuration for no-panic
• Implement floor and ceil in assembly on i586
• Avoid inlining floor into rem_pio2
• Update for new warn-by-default clippy lints
• Use x86_no_sse configuration in more places
• Enable tests that were skipped on PowerPC
• Allow a new lint failure in nightly
• Test building custom targets and resolve an issue probing rustc
• Use the compiler to determine whether or not to enable f16 and f128
• Improved integer utilities, implement shifts and bug fixes for i256 and u256
• Clean up and sort manifest keys
• Fix incorrect result and add tests
• Fix unintentional skips in binop_common
• Use x.biteq(y) rather than x.to_bits() == y.to_bits()
• Fix new dead_code warnings from recent nightlies
• Reuse libm's Caat and CastFrom in compiler-builtins
• Reuse MinInt and Int from libm in compiler-builtins
• Add a note saying why we use frintx rather than frintn
• Typo in README.md
• Update CmpResult to use a pointer-sized return type
• Clean up unused files
• Fix i256::MAX

---

This PR was generated with release-plz.

ci: allow-regressions

[rustbot]

This PR was rebased onto a different main commit. Here's a range-diff highlighting what actually changed.

Rebasing is a normal part of keeping PRs up to date, so no action is needed—this note is just to help reviewers.

[tgross35]

This keeps showing a weird regression:

 icount::icount_bench_fmodf_group::icount_bench_fmodf logspace:setup_fmodf()
Performance has regressed: Instructions (40517 -> 44552) regressed by +9.95878% (>+5.00000%)
  Baselines:                      hardfloat|hardfloat (old)
  Instructions:                       44552|40517                (+9.95878%) [+1.09959x]
  L1 Hits:                            50402|45358                (+11.1204%) [+1.11120x]
  LL Hits:                                3|1                    (+200.000%) [+3.00000x]
  RAM Hits:                              14|13                   (+7.69231%) [+1.07692x]
  Total read+write:                   50419|45372                (+11.1236%) [+1.11124x]
  Estimated Cycles:                   50907|45818                (+11.1070%) [+1.11107x]

Nothing changed here of course. It looks like the difference is running on a AMD EPYC 7763 64-Core Processor (most CI runs) vs. a Intel(R) Xeon(R) Platinum 8370C CPU @ 2.80GHz (this PR). Kind of interesting.