Improve semiring Pareto coverage with focused frontier-behavior tests

[Copilot]

This PR addresses the coverage gap by adding targeted unit tests for Pareto frontier semantics in the semiring subsystem. The change is intentionally narrow: test-only additions for uncovered/under-covered behavior in py3plex.semiring.pareto.

• What changed

  • Added tests/test_semiring_pareto.py to cover:
    • dominance-based rejection in ParetoSet.add()
    • pruning of dominated existing vectors
    • deterministic max_size truncation behavior
    • epsilon-tolerant near-equality handling
    • dimension-mismatch handling through public API behavior
    • union() merge semantics (including duplicate retention behavior in current implementation)
    • cartesian_combine() componentwise aggregation + pruning
    • pareto_semiring_spec() construction (zero, one, plus, times)
    • __repr__ contract

• Design notes

  • Tests validate behavior through public APIs (add, union, cartesian_combine, spec operators), reducing coupling to internals.
  • Existing semantics are documented in assertions (including duplicate handling in union) so future behavior changes are explicit.

from py3plex.semiring.pareto import ParetoSet

frontier = ParetoSet(max_size=2)
frontier.add((2.0, 1.0))
frontier.add((1.0, 2.0))
frontier.add((0.0, 3.0))

assert frontier.to_list() == [(0.0, 3.0), (1.0, 2.0)]