fix: allow equal-prime pairs in primelib Goldbach decomposition

[mohammmad357]

Describe your change

• Add an algorithm?
• Fix a bug or typo in an existing algorithm?
• Add or change doctests? -- Please avoid changing code and tests in one PR.
• Documentation change?

maths.primelib.goldbach accepts even integers greater than two, but its
inner search begins at i + 1, excluding pairs with the same prime twice.
Consequently 4 and 6 reach the final assertion with an empty answer instead of
returning [2, 2] and [3, 3]. Start j at i so equal-prime pairs are
considered. This is a one-line correctness fix; existing doctests are unchanged.
The existing result for 8 remains [3, 5].

Validation and limitations

Base 215c1f0710f610aaf5ebaeab04edacec3f5308cd, independent branch
fix-goldbach-repeated-prime, one algorithm file only. I checked the open
PR list and searched Goldbach issues before preparing this fix; no open PR
addressing this case was found.

On Python 3.13.7, a full import is blocked by unrelated unparenthesized
multi-exception syntax in maths/greatest_common_divisor.py (requires 3.14+).
For local evidence I compiled the unchanged function ASTs of goldbach,
get_prime_numbers, is_prime and is_even, with their real math.sqrt
dependency. No function body was adapted or mocked:

• Before: 4 and 6 each raise the final AssertionError.
• After: every even input from 4 through 500 (249 cases) returns two primes
  with the requested sum; primality was checked with a separate isqrt-based
  trial-division oracle.
• All 19 existing doctests belonging to these four functions pass.
• git diff --check passes.

This isolated check is not the full module or upstream Python 3.15t suite.
Ruff/pre-commit installation timed out downloading from files.pythonhosted.org;
these tools are unavailable locally. Kept as a draft pending upstream CI and
review. Prepared with AI coding assistance, with the test scope disclosed.

Checklist

• I have read CONTRIBUTING.md.
• This pull request is all my own work -- I have not plagiarized.
• I know that pull requests will not be merged if they fail automated tests.
• This PR only changes one algorithm file.
• All new Python files are placed inside an existing directory. (No new files.)
• All filenames are lowercase, without spaces or dashes.
• Functions and variables follow Python naming conventions.
• All function parameters and return values are annotated with type hints.
• All functions have doctests that pass the automated testing.
• All new algorithms include an explanatory URL.
• Open issues resolved by this PR have a closing keyword.

No new algorithm or open issue is claimed. Full upstream automated testing
has not yet run; the scoped local doctest result is above.