[patrickballard]

Description

Fixes #5072.

The ordering operators (<, >, <=, >=) compare two numeric-looking operands as numbers through _safe_compare, but == and != compared the raw values. A numeric string such as captured stdout "10" was therefore both >= 10 and <= 10, yet not == 10.

This PR moves the existing "coerce only when both sides look numeric" step out of _safe_compare into a small _coerce_numeric_pair helper, and routes == and != through a new _safe_equals that uses the same helper. Behavior for non-numeric strings is unchanged: "10x" == 10 is still false and "10x" == "10x" is still true.

One consequence worth calling out: two numeric strings now compare as numbers for equality too ("10" == "10.0" is true), matching how "10" > "9" already compares numerically.

Testing

• Added test_equality_coerces_numeric_strings_like_ordering in tests/test_workflows.py, using the reproduction from the issue plus numeric-string pairs and non-numeric strings.
• Checked operators.py in isolation (it is a leaf module) against the issue's cases and existing ordering cases.
• Ran the full suite with python -m pytest tests -q -n auto (LC_ALL=en_US.UTF-8, macOS, Python via uv sync --extra test): 13953 passed, 446 skipped, 5 failed. The same 5 fail on unmodified main in this environment (test_non_windows_uv_tool_dry_run_still_previews_installer and four test_fetch_keeps_malicious_text_in_one_argument_even_if_validation_is_bypassed cases), so they are unrelated to this change.
• The new test fails on main and passes with this change.

AI Disclosure

• I did not use AI assistance for this contribution
• I did use AI assistance (fill in the disclosure below)

AI disclosure: Implemented with Claude Code using Claude Opus 5.5 in autonomous agent mode, on behalf of @patrickballard: issue selection, code, tests and this PR description were AI-generated. The commit carries an Assisted-by: trailer.

🤖 Generated with Claude Code

[Copilot]

🟢 Approval recommended

No active review issues were identified, although tests were not run during this review.

No open issues 🎉

What changed in this PR

Fixes #5072 by making workflow equality operators use the same numeric-string coercion as ordering operators.

Changes:

• Extracts shared numeric-pair coercion and adds an equality helper.
• Routes == and != through the new helper.
• Adds regression coverage for captured stdout, numeric-string pairs, and non-numeric strings.

The PR reports before-and-after regression evidence; tests were not run during this review.

File	Description
tests/​test_workflows.py	Adds equality-coercion regression tests.
src/​specify_cli/​workflows/​expressions/​operators.py	Shares numeric coercion between equality and ordering.
src/​specify_cli/​workflows/​expressions/​engine.py	Uses the equality helper for == and !=.

_{Review effort: Balanced (auto)}