fix(presets): tolerate non-string and non-list catalog fields in preset search/info

[Noor-ul-ain001]

Problem

preset search and preset info crash with a raw Python traceback on catalog payloads that are valid YAML/JSON but not string-typed. Catalog files are user-editable, so these shapes reach the code unvalidated.

PresetCatalog.search had three unguarded assumptions:

Input	Result before this PR
author: 789 + --author 789	AttributeError: 'int' object has no attribute 'lower'
name: 123 / description: 456 + a query	TypeError: sequence item 0: expected str instance, int found
tags: 5 (scalar, not a list) + --tag ci	TypeError: 'int' object is not iterable

Reproduced directly against PresetCatalog.search:

author filter: AttributeError: 'int' object has no attribute 'lower'
query filter: TypeError: sequence item 0: expected str instance, int found
tag filter:   TypeError: 'int' object is not iterable

#3743 fixed only the non-string elements of tags in this method. A non-list tags, and the author / name / description fields, were still unguarded.

Fix

The sibling catalogs already handle every one of these cases — extensions/__init__.py::search and integrations/catalog.py::search coerce with str(...) and gate on isinstance(raw_tags, list). This PR aligns presets with that existing reference pattern rather than inventing a new one.

The same scalar-tags crash also reached the four display sites in presets/_commands.py, which now gate on isinstance(tags, list), matching integrations/_query_commands.py:331.

That includes the local-manifest branch of preset info: PresetManifest.tags returns self.data.get("tags", []) and manifest validation does not enforce list-ness, so a local preset.yaml containing tags: 5 validates successfully and then crashed the display — the same validate-accepts / render-crashes split seen in prior fixes.

While here: preset search printed tags unescaped, so a tag containing [bold] was silently swallowed as a Rich style tag. It now routes through _escape_markup, like the preset list line directly above it.

Tests

Five regression tests added to TestPresetTagsNonString, all driving the full CLI path via CliRunner (testing only the display helper would have missed the backend search crashes):

• test_search_by_author_tolerates_non_string_author
• test_search_query_tolerates_non_string_name_and_description
• test_search_tolerates_non_list_tags
• test_info_tolerates_non_list_tags
• test_search_escapes_rich_markup_in_tags

Each fails before the fix with the exact exception it targets, verified by stashing the source changes:

FAILED ... test_search_by_author_tolerates_non_string_author
  → AttributeError("'int' object has no attribute 'lower'")
FAILED ... test_search_query_tolerates_non_string_name_and_description
  → TypeError('sequence item 0: expected str instance, int found')
FAILED ... test_search_tolerates_non_list_tags
  → TypeError("'int' object is not iterable")
FAILED ... test_info_tolerates_non_list_tags
  → TypeError("'int' object is not iterable")
FAILED ... test_search_escapes_rich_markup_in_tags
5 failed, 2 passed

tests/test_presets.py after the fix: 462 passed, 7 failed. The 7 failures are identical on clean main (457 passed, 7 failed) — pre-existing TestPresetSkills symlink cases that need elevation on Windows. Net effect is +5 passing, no new failures.

🤖 Generated with Claude Code