Use sensible defaults for BarRenderer.bar_value

[ianthomas23]

Describe your changes

This PR adds some sensible defaults for scales used in BarRenderer.bar_value so that they can be omitted by the user. If min or max arguments of a DateScale, LinearScale or LogScale are omitted, they are determined from the data in the corresponding column. An example from some of the example notebooks is that the previous use of

BarRenderer(bar_value=DateScale(min=df[\"Dates\"][0], max=df[\"Dates\"][n - 1]))

can be replaced by

BarRenderer(bar_value=DateScale())

The calculations occur in the DataGrid constructor on the Python side.

Also, bar_value now accepts None and will create either a DateScale or LinearScale depending on the data type of the corresponding column. bar_value already has a default of 0.0 and we cannot change this default without causing a backwards incompatibility, hence the use of None.

This new functionality is not supported in default_renderer, header_renderer, or corner_renderer as they are not attached to a particular data column.

Testing performed

• Manual testing locally.
• New python test test_bar_renderer_defaults added.
• Playwright tests include notebook changes to demonstrate the new functionality in the frontend.

Possible future work

This work could be extended to cover other Renderer classes, and renderers that refer to multiple columns (e.g. default_renderer.