Several minor bugs in plot_trajectory

[niksirbi]

This issue highlights two related problems with the current plot_trajectory function: ambiguous or incorrect plot titles, and mismatched time/space units. Some of this may be also relevant for the plot_occupancy function that's coming with #403.

1. Misleading Titles

When plotting multiple trajectories on the same axis (as demonstrated here), the final plot title is determined by the last individual plotted. This can lead to confusion when multiple trajectories are shown.

In certain other edge cases, the title ends up displaying "Trajectory of None".

Proposed Fix
Rather than trying to infer a suitable title based on the plotted data, we should either:

• Omit the title by default.
• Use a generic title that's always true (e.g., "Trajectory").

Users can always override the title themselves if needed.

2. Erroneous Time/Space Units

The plot_trajectory() function attempts to read space_unit and time_unit attributes directly from the position DataArray. However, these attributes are only set at the Dataset level, meaning plot_trajectory() defaults to "pixels" and "frames" even if the actual units are, for example, "meters" and "seconds".

Possible Solutions

1. Pass the entire Dataset
   This ensures that plotting functions have access to dataset-level attributes. However, it complicates matters when a dataset contains multiple variables (e.g., position, confidence) and forces users to carry unneeded data.

2. Inherit attributes in the DataArray
   On dataset creation or file loading, propagate relevant attributes (e.g., time_unit, space_unit) to each DataArray. This requires deeper changes in the I/O layer and consideration of broader package implications.

3. Stop attempting to infer units
   As a quick hotfix, simply label plot axes as "x", "y", and "Time" without specifying units. This avoids misleading labels until we have a more comprehensive solution.

Recommendation
I propose implementing Option 3 for now, allowing a functional v0.0.24 release. A more complete fix (Option 2) can be pursued later once we have time to address the underlying I/O and attribute-inheritance logic.

[niksirbi]

I'm probably the one to blame fro the 2nd bug, because I encouraged relying on space_unit and time_unit to begin with, not realising that these are not defined on the position data array level.

[stellaprins]

Thanks @niksirbi. Used your suggestions and double checked the plot_occupancy function that's coming with #403 but this doesn't seem to have the issues.

I did not see plot_occupancy used in any of the examples and was tempted to add a plot, like the one below, to Express 2D vectors in polar coordinates (but it's probably better to have a seperate plot example than to bloat the current examples!).

[niksirbi]

I did not see plot_occupancy used in any of the examples and was tempted to add a plot, like the one below, to Express 2D vectors in polar coordinates (but it's probably better to have a seperate plot example than to bloat the current examples!).

I am currently working on updating the polar coordinate example #317 , and I also want to tackle #410 in combination with #415, which should take care of showcasing the occupancy plot. So don't worry about it.