Reject unstable direct biquad coefficients

[be-student]

What & why

Fixes #285 by rejecting directly supplied feedback weights unless their poles are strictly inside the unit circle. This preserves the invariant that every accepted BiquadCoefficients value is stable, so Biquad::settle_to cannot divide by a zero steady-state denominator and seed non-finite state.

Tests cover the reported zero-divisor pair, other unstable boundary cases, and unchanged settling for designed low-pass and notch filters. The changelog and public API documentation describe the new validation. This contribution was implemented with AI assistance and reviewed through the repository's test and lint gates.

Checklist

• cargo test + cargo clippy --all-targets clean locally
• New public APIs have a doc example (no new public API)
• No unwrap/expect/panic on library paths (typed errors instead)

[github-actions]

Thanks for the first PR! 🎉 You'll get a first review within 48 h. CI runs the 5-target matrix — a budget-gate failure isn't your fault, see ci/README.md.