fix: use generic security level in aux precompute instead of hardcoded SecurityLevel128

[Mandalorian7773]

Closes #185

Problem

get_aux_inner in tests/src/lib.rs calls precompute_multiexp_table with a hardcoded SecurityLevel128, ignoring the generic L parameter already in scope on the function signature.

This causes all secp384r1 tests to fail because that curve uses SecurityLevel192 (via the CurveParams trait impl at line 509), creating a type mismatch during aux info precomputation. The bug was silent because CI runs cargo test --release, which compiles out all debug_assert! calls — so no assertion fired to catch it.

Fix

1. One-line bug fix (tests/src/lib.rs):

-  .precompute_multiexp_table::<cggmp24::security_level::SecurityLevel128>()
+  .precompute_multiexp_table::<L>()

2. Enable debug assertions in release test builds (Cargo.toml):

Added [profile.release.package.*] overrides so debug_assert! calls inside workspace library code fire during cargo test --release, as CI does. Third-party dependencies are unaffected.

[profile.release.package.cggmp24]
debug-assertions = true

[profile.release.package.cggmp24-keygen]
debug-assertions = true

[profile.release.package.key-share]
debug-assertions = true

[profile.release.package.paillier-zk]
debug-assertions = true

This ensures existing debug_assert! guards in paillier-zk, key-share, cggmp24-keygen, and cggmp24/src/signing.rs will catch invariant violations in future CI runs.

Impact

• Fixes all secp384r1 signing tests
• No behavioral change for secp256k1, secp256r1, or stark
• Prevents silent regressions of this class in CI going forward

---

[Copilot]

Pull request overview

Fixes aux-info multiexponentiation precomputation in tests so it respects the generic security level L (instead of always using SecurityLevel128), resolving type/parameter mismatches for curves like secp384r1 that require SecurityLevel192.

Changes:

• Replace hardcoded SecurityLevel128 with the generic L in get_aux_inner when calling precompute_multiexp_table.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Mandalorian7773]

Hi @survived , this PR fixes the secp384r1 failures.
All tests pass locally, would love your review.

[maurges]

Nitpick: it's evident that crates not listed here are not affected. I'd rather have a comment like « the binaries using the listed crates are not affected since these options only affect the builds in the context of this workspace », except I phrased it terribly. This fact was not evident to me

[maurges]

Looks good, but your patch fails the cargo fmt check

[Mandalorian7773]

Fixed @maurges

[maurges]

Looks good otherwise

[maurges]

Replace with:

Enable debug assertions in the release builds since we have to run the tests in release mode, because they are too slow

A comment should rarely be an explaination, it should state intent and reason.

And remove the lines between sections, they make them seem unrelated