Make det queries discover determinant operations

[yuelgrace1810-ops]

Problem

On current main, the standard mathematical abbreviation det ranks matrix characteristic-polynomial operations first because their descriptions contain det(tI-A). The actual rational and symbolic determinant operations do not mention the abbreviation in their searchable metadata.

This is a demonstrated leaf-level vocabulary gap. It follows the maintainer direction recorded on #1059: keep discovery lexical and fix concrete terminology gaps in the owning MathTool metadata instead of introducing a new retrieval architecture.

Change

• Add the standard det abbreviation to the titles of the rational and symbolic determinant operations.
• Add a held-out catalog regression asserting that both determinant operations rank ahead of the unrelated characteristic-polynomial matches.

The property-style regression intentionally does not impose an order between the two legitimate determinant operations, addressing the maintainer's review concern about future catalog additions.

This changes discovery metadata only. Request/result models, mathematical implementations, bounds, and execution semantics are unchanged.

Validation

• Rebased onto current main after merged Fix references to the moved agent evaluation guide #2029.
• make check — 1,309 passed; Ruff, formatting, complexity, mypy, and ordinary owner tests passed.
• Focused catalog regression suite — 16 passed.
• make test-integration TESTS=tests/integration/catalog/test_builtin_examples.py — 239 passed on the original validated tree; the rebase changed only the unrelated moved-guide paths.
• git diff --check — passed.

Scope

This PR intentionally does not change global tokenization or ranking semantics and does not attempt to solve numeric-literal query interpretation.

[morluto]

Review verdict: no correctness blocker found

This is a narrow, appropriate discovery-metadata change. Adding det as a standalone title token makes the existing token-boundary search discover the numeric and symbolic determinant operations without changing either mathematical implementation or public schemas.

The regression checks the user-visible ranking rather than an internal score, which is the right level. One nonblocking concern is brittleness: asserting the exact first two operations couples the test to future catalog additions that may legitimately also be strong det matches. A property-style assertion that both determinant operations precede unrelated matches would preserve the intent with less ranking-order coupling, but the current behavior is coherent.

[yuelgrace1810-ops]

Addressed the review observation: the regression now checks that both determinant operations precede both unrelated characteristic-polynomial matches, without fixing the order between legitimate determinant operations. The branch is rebased onto current main and make check passes all 1,309 tests.