Improve quiver API to improve clarity and consistency

[JamesVarndell]

Is your feature request related to a problem? Please describe.

Matplotlib's quiver API is not very well-suited to plotting meteorological data. Calibrating arrow sizes can only be done based on a combination of the axis dimensions, the axis units (e.g. lat & lon), size in inches or the resolution of the screen.

Instead, this issue proposes an additional method of sizing quiver arrows, based on the density of arrows on the plot and a scale value based on the input data's units.

Describe the solution you'd like

We should introduce new arguments for scale_units in quiver:

• density - default, sizes based on the sparsest density (x or y)
• density-y - sizes based on the density of qrrows in y
• density-x - sizes based on the density of qrrows in x

In addition, in combination with these values the scale argument should accept statistics like mean, max, median etc.

Describe alternatives you've considered

No response

Additional context

No response

Organisation

ECMWF

[JamesVarndell]

Moved to version 0.6.0. The issue with quiver plotting being incorrect on the PlateCaree projection is now solved, but adding this density API is too complex to finish for version 0.4.