Restructure edit section in GUI guide

[lochhh]

This PR attempts at an "extendable" structure to make it easier for us to document new edit types with the expected behaviour/consequences.

Summary of changes:

• creates new section labels (e.g. (target-points-layer)=, (target-load-tracked-data)=) to refer to sections within the same doc
• introduces the GUI elements involved in editing (viewer, Edit tracked data menu, edit timeline, Save tracked data menu) up front, before any steps
• splits the old "Move and remove keypoints" steps into Select keypoints (common to all edits) and Edit types, with one subsection per edit type (currently Move and Remove); new edit types (e.g. add keypoints, swap keypoint labels, swap identities) can be added as further subsections
• states behaviour common to all edits (menu expands, timeline appears, tracks layer stays in sync) once in the Edit types intro - no need to repeat per edit type
• folds "Resume an editing session" into Save edits, and replaces the "Using edited data in Python" dropdown with a short tip linking to the netCDF section of the I/O guide
• tries to keep the wording future-proof where features are expected to grow
• corrects details against the implementation: transposing the viewer does not disable editing (only moving time off the frame slider or switching to 3D does); the timeline spans all frames of the dataset rather than the "recording"; edited is False (not NaN) for unedited keypoints
• turns the bboxes limitation into a warning and moves it next to the intro
• renames "Load the tracked dataset" to "Load tracked data" to match the menu name
• removes the GIFs that no longer match the new structure and leaves <!-- TODO --> placeholders describing the GIFs/screenshots still needed

[anna-teruel]

All changes look good to me! Thank you @lochhh for taking the time to review my text. Appreciate it. I will merge this PR with my PR push it to #1126