feat: add pure-Python results post-processing helpers

[scarwizz]

Summary

Closes #57

This PR adds pure-Python, NumPy-free results post-processing helpers (probabilities_to_counts, relabel_to_bitstrings, marginal, and expectation_z) to ionq_core/results.py. These functions operate directly on standard Mapping[str, float] dictionaries to support downstream SDKs without adding heavyweight dependencies.

Key Features & Implementation Details

• NumPy-free Logic: All implementations use pure Python built-ins.
• probabilities_to_counts: Implements the largest-remainder method (Hare quota) to guarantee counts sum exactly to the requested shots, avoiding floating-point drift. It includes a deterministic tie-breaker (sorting by state integer ascending) for edge cases where remainders are identical. Gracefully handles shots=0 and empty mappings.
• relabel_to_bitstrings: Converts integer keys to zero-padded bitstrings using big-endian ordering (qubit 0 is the most significant bit), matching IonQ's Cloud API conventions.
• marginal: Safely computes marginal distributions while maintaining the requested ordering of the subset qubits. Safely handles duplicate qubits and negative indices.
• expectation_z: Calculates Z-parity expectation using highly efficient, native Python int.bit_count().
• Strict Validation: The _validate_probabilities helper rigorously checks for and rejects invalid probabilities (NaN, ±Inf, negative values), out-of-bound state integers, and invalid qubit indices.
• Repository Integration: Helpers are properly re-exported in ionq_core/__init__.py. The Jinja template was updated to prevent CI failures, NumPy-free conventions were added to AGENTS.md, and a CHANGELOG.md entry is included. Proper SPDX headers are present in all new files.

Test Plan

Achieved 100% branch coverage on ionq_core/results.py, satisfying the repository's strict CI requirements.

Tests include:

• 39 comprehensive tests in tests/test_results.py.
• Real simulator response fixtures (two-qubit Bell state: {"0": 0.5, "3": 0.5}, three-qubit GHZ state).
• Deterministic rounding verification for complex fractional splits.
• Full edge-case coverage for invalid inputs, missing fields, and out-of-bounds parameters.

Local checks passed successfully:

• uv run pytest (100% coverage)
• uv run ruff check (0 errors)
• uv run ruff format --check
• uv run ty check ionq_core/

unitaryHACK AI Disclosure:
I used Gemini to help draft the pure-Python logic, implement algorithms, format code to repository standards, and write comprehensive edge-case unit tests. I then manually ran the code, debugged local execution environment errors, and verified the logic locally using the uv run pytest and ruff toolchains to ensure 100% branch coverage and deterministic execution.

---

Important

Most code in ionq_core/ is auto-generated and overwritten on regeneration.
See CONTRIBUTING.md for which files are safe to edit.