Add support of curves with larger scalar sizes

[survived]

Currently, the crate only supports curves with scalars of 32 bytes. We want to expand it to work with curves that have scalar size $\ge 32$ bytes, e.g. curve448, P-384, P-521, and so on.

The size of the scalar has been hardcoded into library in several places, therefore if we simply add a curve with larger scalar size, library will break in some places causing runtime panics.

Expected development path:

1. Add curves with different scalar sizes to the library, update all tests to use these curves, and observe what tests get broken
2. Fix the issues until tests pass
3. Update cggmp24 and givre crates to support these curves, update all tests to use these curves and make sure all tests pass
4. Carefully review the library to identify any other places where we could make assumptions on the scalar size

Once all of this is done, we can release a new version of generic-ec[-curves] with the added curves. Preferably, that would be a release without any breaking changes in API unless we absolutely have to break the API.

Related work:

• Fix hardcoded scalar size in multiscalar multiplication Fix hardcoded scalar size in Straus NAF computation #57

If you're willing to take part in development, reach out to us in discord #lockness-contribute channel.

[sridhar-panigrahi]

Hey @survived , I've started working on this! Opened #59 which adds secp384r1 (P-384)
as the first curve with a larger scalar size (48 bytes). All existing
tests are instantiated for the new curve and pass — no breakage found
beyond the Straus NAF fix already in #57.

Happy to continue with the next steps from the development path — adding
more curves with different scalar sizes and then updating cggmp24 and
givre.

[survived]

Hi @sridhar-panigrahi, thanks for your help in this issue! I've released generic-ec with the secp384r1 curve and NAF changes you made.

What concerns this PR, the library could still benefit from other "big" curves such as curve448 (so we can integrate it into givre), and P-521. I leave issue open for anyone to pick it.

[sridhar-panigrahi]

@survived That's awesome, thanks for getting it released! I'm definitely interested in continuing with this — curve448 sounds like a good next step since it'd help with givre. I'll dig into it and open a PR when it's ready. P-521 can follow after that.

[sridhar-panigrahi]

Hey @survived, just opened #61 adding Curve448 support! Used
ed448-goldilocks-plus as the backend. All tests are passing. Let me know
if anything needs adjusting — happy to pick up P-521 once this one lands.
Thanks!