[caglarpir]

Problem

Interval sampling maps every video stream to one output (-map 0:v), and ffmpeg numbers the frames of all of them in one sequence. A GoPro MAX .360 video has two video streams, so frame i, taken at ⌊(i-1)/2⌋ intervals, is stamped at (i-1) intervals. On a 5-minute normal-speed .360 sampled every second, video_process gives 588 samples. 298 of them fail, and the other 290 are placed a median of 73 m (max 190 m) from where they were taken, with --geotag_source gpx and with --geotag_source gopro alike.

Changes

Interval sampling samples the video stream with the highest resolution, as distance sampling does. A video with a single video stream, which is nearly every video, is sampled as before, and its samples keep their _v_ names. A video with several streams gets the index of the sampled stream in the names, as distance samples do.

Verification

• The 5-minute .360: 294 samples, 290 geotagged with gpx at a median of 2 m (the GPX's own error), 293 with gopro at 0 m.
• Normal single-stream videos (GoPro HERO7 and HERO8, BlackVue, Insta360 Pro), interval sampling with gpx and with their own source: output identical to main.
• Unit: 914 passed, 18 skipped, 1 xfailed. Integration: 69 passed. mypy, ruff check, ruff format --check and usort diff are clean.

With #852: one hunk in _sample_single_video_by_interval: probe once, then #852's telemetry and rate.