[caglarpir]

A GPX track that misses a video hands the video on to the next geotag source. When no later source geotagged it, the error reported was that of the last source to fail, which hid the miss. A GoPro video whose own GPS is too noisy, given a GPX of another day, failed as "GPS is too noisy": the very problem the GPX was given to work around. The Desktop Uploader, whose chain runs gpx, gpx, exiftool_xml, native and exiftool_runtime, shows only that error.

After a GPX that misses the video	Before	After
a later source fails as too noisy, no GPS found, file not found or ExifTool not found	that failure	MapillaryOutsideGPXTrackError, the miss
a later source fails with an error about the file itself, such as MapillaryStationaryVideoError	that error	unchanged
a later source geotags the video	a warning about the miss	unchanged

The error, for a noisy GoPro video given a GPX recorded months before:

The GPX track in GS018205.gpx (2026-05-29T12:47:06+00:00 to 2026-05-29T12:53:00+00:00) misses the video GS018205.360 (2026-09-11T15:39:40.599000+00:00 to 2026-09-11T15:39:56.099000+00:00) by 105.1 days. Check that the GPX file belongs to this video, then the camera clock and the time zone of the GPX timestamps

Tests: unit tests for the Desktop Uploader's chain, for each kind of later failure, for a later error that ends the chain, and for a later source that geotags the video. Also checked by hand through process with the Desktop Uploader's arguments, on a noisy GoPro video with a GPX of another day, of the same day 3 hours off, and of the video itself.

First of three stacked PRs: this one, then #939 and #940. None of them touches README.md or docs/; the documentation follows in its own PR.