perf: use NAF decomposition in mulWindow

[yelhousni]

Description

Use ecc.NafDecomposition in utils.go in mulWindowed.

Fixes #132

Type of change

• New feature (non-breaking change which adds functionality)

How has this been tested?

Same tests pass.

How has this been benchmarked?

e.g. BLS12-377:

cpu: AMD EPYC 9R14
BenchmarkG1AffineScalarMultiplication/method=GLV-wNAF/w=5-32     129970        115066        -11.47%
BenchmarkG2AffineScalarMultiplication/method=GLV-wNAF/w=5-32     405302        333270        -17.77%

Checklist:

• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have added tests that prove my fix is effective or that my feature works
• I did not modify files generated from templates
• golangci-lint does not output errors locally
• New and existing unit tests pass locally with my changes
• Any dependent changes have been merged and published in downstream modules

---

Note

Implements NAF-based scalar multiplication across all curves and updates codegen.

• Replaces old 2-bit windowed double-and-add with NAF-based mulWindowed, adding mulWindowedMixed (affine input) and an affine wrapper; handles zero/negative scalars and reduces scalars mod fr where needed
• Adjusts ScalarMultiplication/ScalarMultiplicationBase to use new paths and return early, using affine generators (*GenAff) in non-GLV cases
• Minor flow tweaks (e.g., moving FromAffine only when needed) and identical changes applied to G1/G2 for bls12-377, bls12-381, bls24-315/317, bn254, bw6-633/761, grumpkin, secp256k1
• Updates internal/generator/ecc/template/point.go.tmpl to generate the new logic

^{Written by Cursor Bugbot for commit 833801b. This will update automatically on new commits. Configure here.}

[gbotrel]

nice would be nice if we could have a version with no big int for the naf decomposition to avoid allocs (i.e. if scalar < modulus, use the non-big-int version) ?

[yelhousni]

nice would be nice if we could have a version with no big int for the naf decomposition to avoid allocs (i.e. if scalar < modulus, use the non-big-int version) ?

The scalar decomposition can output negative integers which would be mod-reduced and the bit-length would increase beyond the |mod|/2 bound.