Investigate solver options for finding canopy layer closure heights

[davidorme]

The closure heights of the canopy layers in pyrealm.demography.canopy.Canopy are found by using a root solver:

pyrealm/pyrealm/demography/canopy.py

Lines 69 to 86 in d4f2426

	# TODO - the solution here is predictably closer to the upper bracket, might
	# be a better algorithm to find the root.
	solution = root_scalar(
	solve_community_projected_canopy_area,
	args=(
	community.cohort_data["stem_height"].to_numpy(),
	community.cohort_data["crown_area"].to_numpy(),
	community.cohort_data["m"].to_numpy(),
	community.cohort_data["n"].to_numpy(),
	community.cohort_data["q_m"].to_numpy(),
	community.cohort_data["canopy_z_max"].to_numpy(),
	community.cohort_data["n_individuals"].to_numpy(),
	target_area,
	False, # validate
	),
	bracket=(0, starting_guess),
	xtol=layer_tolerance,
	)

The default method works well, but we might be able to improve the performance if we can make use of the fact that the root is likely to be much closer to the upper bound (the previously closed layer) than the lower bound (the ground).

[davidorme]

Actually, it might be more effective to set the lower bound - we know the number of stems and crown areas of each cohort and the vertical height of maximum crown area ($z_{max}$), so can determine some known heights where a particular community crown area has to be filled.

It's probably a tiny runtime saving but in simulation this routine will be called a lot.