Add filename pattern cross-validation grouping

[gbeane]

Adds a third leave-one-group-out cross-validation grouping option, Filename Pattern, alongside the existing Individual Animal and Video options (configured in Project Settings → Cross-Validation).

What it does

• A regular expression is applied to each video filename; videos that produce the same key form one CV group. If the pattern has a capture group, the captured text is the key (cage_(\d+) → 0042); otherwise the whole match is used. Files that don't match are each placed in their own group.
• New cv_grouping_regex project setting (stored in project.json); only affects new training runs.

UI

• Regex field shown only for the Filename Pattern strategy, with live validation that blocks saving an empty or invalid pattern.
• Live preview: a summary line plus a collapsible breakdown of how the project's videos partition into groups, with videos excluded from training marked.
• Inline help updated with a cage-ID example.

Plumbing

• Grouping implemented in Project._assign_cv_group_ids; the regex is threaded through the binary/multiclass Train-button gating counters, the training report, and the CLI.
• Group-mapping consumers (CV result labels, training export, excluded-video logic) updated to handle groups that span multiple videos.

Docs & tests

• User guide updated in both copies (online docs/ and in-app help).
• New tests for the helpers, group assignment, threshold counters, settings persistence, and the dialog preview/validation. Full suite green (796 tests); lint and format clean.

[Copilot]

Pull request overview

Adds a new leave-one-group-out cross-validation grouping strategy, Filename Pattern, allowing users to group videos into CV folds based on a regex-derived key from filenames. This extends the existing Individual Animal and Video grouping modes and threads the new setting through UI, project plumbing, reporting, CLI, export, docs, and tests.

Changes:

• Add FILENAME_PATTERN grouping strategy plus core helpers (compile_grouping_regex, filename_group_key) and a persisted project setting cv_grouping_regex.
• Implement filename-pattern grouping in Project._assign_cv_group_ids and update downstream consumers (excluded-video handling, CV labels, threshold gating, exports, reports, CLI).
• Add UI controls with inline regex validation and a live “group preview”, plus comprehensive tests and doc updates (both in-app and online copies).

Reviewed changes

Copilot reviewed 26 out of 26 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
tests/ui/test_settings_dialog.py	Adds UI tests for strategy/regex roundtrip, validation behavior, and preview rendering/visibility.
tests/ui/_fakes.py	Extends UI fakes to include cv_grouping_regex.
tests/project/test_settings_manager.py	Verifies defaulting and persistence reading for cv_grouping_regex.
tests/project/test_cv_grouping.py	Adds focused tests for Project._assign_cv_group_ids under filename-pattern grouping.
tests/classifier/test_multi_class_classifier.py	Tests multi-class threshold counting behavior under filename-pattern grouping (including invalid regex).
tests/classifier/test_classifier.py	Tests binary threshold counting and gating under filename-pattern grouping (including invalid regex).
src/jabs/ui/training_thread.py	Threads cv_grouping_regex into training report construction.
src/jabs/ui/training_strategy.py	Extends report-building API to carry cv_grouping_regex into TrainingReportData.
src/jabs/ui/settings_dialog/settings_group.py	Adds a validate() hook to settings groups to block saving invalid input.
src/jabs/ui/settings_dialog/settings_dialog.py	Adds “validate all groups before save” behavior and passes project videos into the CV settings group for preview.
src/jabs/ui/settings_dialog/cross_validation_settings_group.py	Implements regex field visibility, inline validation, debounced live preview, and updated help text.
src/jabs/ui/main_window/central_widget.py	Threads cv_grouping_regex into train-button gating counters for both binary and multi-class.
src/jabs/scripts/cli/cross_validation.py	Includes cv_grouping_regex in CLI-generated training reports.
src/jabs/resources/docs/user_guide/gui.md	Updates in-app user guide with Filename Pattern explanation and example.
src/jabs/project/settings_manager.py	Adds cv_grouping_regex property reading from project settings.
src/jabs/project/project.py	Implements filename-pattern group assignment; updates excluded-group logic; threads regex into feature extraction grouping.
src/jabs/project/export_training.py	Updates training export to store group labels appropriately for filename-pattern groups.
src/jabs/classifier/training_report.py	Records cv_grouping_regex in markdown and JSON report outputs.
src/jabs/classifier/multi_class_classifier.py	Adds filename-pattern aggregation support to multi-class threshold counting/gating.
src/jabs/classifier/cross_validation.py	Updates CV test-group label rendering to prefer regex label when present.
src/jabs/classifier/classifier.py	Adds filename-pattern aggregation support to binary threshold counting/gating.
packages/jabs-core/tests/test_cv_grouping.py	Adds unit tests for the new enum member and helper functions.
packages/jabs-core/src/jabs/core/enums/cv_grouping.py	Adds FILENAME_PATTERN and regex helper utilities.
packages/jabs-core/src/jabs/core/enums/init.py	Re-exports new helpers for consumers.
packages/jabs-core/src/jabs/core/constants.py	Adds CV_GROUPING_REGEX_KEY constant.
docs/user-guide/gui.md	Updates online user guide with Filename Pattern explanation and example.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.