ggplot2 implementation of time series & trends functions

[jack-davison]

This PR implements ggplot2 versions of 'time series and trends' functions, as part of the drive to provide a modern API for openair.

Tracking implementation:

• trendLevel() -> plot_heatmap()
• calendarPlot() -> plot_calendar()
• timeProp() -> plot_trend_bars()
• timePlot() -> plot_trend_lines()
• smoothTrend() -> plot_trend_smooth()
• TheilSen() -> plot_trend_theilsen()
• timeVariation() -> plot_variation()

This PR also adds some necessary structural changes to initialise ggplot2 in openair:

• Adds ggplot2 (>=4.0.0) and scales as dependencies.
• Adds exported "options" functions to help manage options in a consistent way:
  • discretise_opts for flexibly discretising a scale (e.g., plot_calendar, plot_heatmap)
  • facet_opts for controlling how facets are handled
  • (soon) windflow_opts for controlling the appearance of windflows
• Adds internal functions:
  • facet_wd() for wind direction faceting. I think this could be exported one day.
  • layer_windflow() for adding 'windflow' arrows (e.g., in plot_calendar, plot_trend_line)

Old functions are still present, and documentation cross-links them.

There are a few initial benefits to the ggplot2 functions:

• ggplot2 handles a lot of things better than lattice, meaning the functions are a lot shorter.
• It is much easier to have type be 2 things and create a grid.
• There are some initial 'easy wins' - e.g., adding windflow to plot_heatmap() (which fixes a very old issue - [Feature request]: wind vectors in trendlevel() #49)

Some Examples: