perf: optimize SepticExtension field arithmetic operations

[Lil-Duckling-22]

Motivation

The previous implementation of polynomial multiplication in SepticExtension performed unnecessary clone operations and computed results involving zero coefficients. This introduced avoidable overhead, especially in arithmetic-heavy workloads.

Solution

Optimized the Mul implementations for both owned and borrowed Polynomial:

• Skips multiplication by zero coefficients, significantly improving performance in sparse polynomials.
• Uses unsafe indexing (get_unchecked_mut) in the inner loop to reduce bounds checks in a performance-critical path. This is safe because index bounds are guaranteed by construction.
• Reduced redundant allocations and cloning of coefficient values.

PR Checklist

• Added Tests
• Added Documentation
• Breaking changes

[leruaa]

As mentioned in the pull request template, trivial PRs are generally not accepted to not encourage PR farming.