[m-messer]

Stacked on #300, which is stacked on #299 and the rc4 pin in #298. Hold until #298 merges. Replaces #276.

The parsing is in lambda-feedback/lambda-feedback-packages#13. <, <=, >, >= and != (and ≠, ≤, ≥) now parse to SymPy relations. 1 < x < 5 parses to an And of two relations. Grading, feedback, preview and docs stay here.

Behaviour (unchanged from #276)

Answer	Accepted	Rejected
2x - 10 >= 0	x >= 5, 5 <= x, 4x - 20 >= 0, 10 - 2x <= 0	x > 5 (strictness), x <= 5 (opposite direction)
1 < x < 5	5 > x > 1, 0 < x - 1 < 4, 2 < 2x < 10	1 <= x < 5 (strictness)
x != 5	5 != x, 2x != 10, x - 5 != 0, x ≠ 5	x = 5

Where it goes

• New app/context/symbolic/inequalities.py: inequality_bounds and check_inequality_equivalence, along with the single, chained and != helpers. The result is an InequalityComparison enum, and each member's name is the criterion tag suffix (_TRUE, _WRONG_DIRECTION, _STRICTNESS_MISMATCH, ...). This replaces Support inequalities in the answer and response #276's mix of True/False/None/str and its result-to-tag dict. Support inequalities in the answer and response #276's bare except Exception is now except TypeError.
• check_equality (app/context/symbolic/comparison.py): relations are caught before the lhs - rhs subtraction, which raised a TypeError on them.
• app/context/symbolic/graphs/equivalence.py: add_inequality_equivalence_nodes. criterion_equality_node uses it after set equivalence and before equalities.
• app/feedback/symbolic.py: the seven new INTERNAL strings.
• app/preview/symbolic.py: the response is split only on a lone =, so x >= 5 and x != 5 preview as one relation.
• Docs in app/docs/user.md and app/docs/dev.md.

Not changed: check_order still subtracts lhs - rhs without a guard. #276 didn't touch it either.

Tests

• Ported Support inequalities in the answer and response #276's evaluation tests into symbolic_evaluation_test.py, and its preview test into symbolic_preview_test.py. The parser tests are in packages#13.
• 1546 passed, 1 failed: test_radians_to_frequency, which fails before this PR too. The two a!=b factorial cases that fail on Pin the packages to the rc4 preview #298–Add significant figures and decimal places parameters #300 now pass.
• mypy is clean. The blocking flake8 selection reports 0. In the warning-only run, check_equality complexity rises to 13, and _compare_single_inequality is 13.

🤖 Generated with Claude Code