[gbradham]

Problem

A folder of videos run through Batch Pose (each video beside its …DLC_<model>.csv, _raw companion and _zones folder) could not be opened in Session Review:

• Open recording folder reported that it "holds no GLIDER recording … and no ethogram_raw.csv".
• Open folder as cohort found no sessions, because a bare video only counted as a session if the pup pass had left a _pups.npz beside it.

Fix

• discover_sessions now treats a video as a session when find_pose_csv finds its pose CSV, the same way pup-tracked videos already work. A video that already has a session is skipped so it is not listed twice: an ethogram scored from it, its pup session, or the GLIDER recording that filmed it (whose tracking the recording session already reads). Behaviour for existing cohorts is unchanged.
• open_path: when a folder is not one recording and does not hold exactly one ethogram, it now opens every session it finds as a cohort, instead of showing the error. The error still appears when nothing is found, or when the folder is a recording that failed to load.
• The empty-cohort message now also mentions videos with pose or pup tracking.

Verification

• New tests: test_a_video_tracked_by_batch_pose_is_a_session, test_a_pose_tracked_video_with_another_session_is_not_listed_twice, and TestOpenVideo::test_a_folder_of_pose_tracked_videos_opens_them_all.
• tests/unit/analysis and tests/unit/gui/behavior: 2183 passed. ruff and black are clean.
• On a real folder of 7 rat open-field videos on the lab share, Open recording folder loaded all 7 as a cohort in 2.3 s with no error dialogs. Each session got its own DLC_train-18.csv and its full length (675–695 s).