Unify default wavespeed estimate for flux_lax_friedrichs and StepsizeCallback

[DanielDoehring]

As mentioned in #2233 , we are currently not consistent how the default wave speed of the (by far most commonly used) numerical flux flux_lax_friedrichs

Trixi.jl/src/equations/numerical_fluxes.jl

Lines 209 to 222 in e93de2a

	function FluxLaxFriedrichs(max_abs_speed = max_abs_speed_naive)
	FluxPlusDissipation(flux_central, DissipationLocalLaxFriedrichs(max_abs_speed))
	end
	function Base.show(io::IO, f::FluxLaxFriedrichs)
	print(io, "FluxLaxFriedrichs(", f.dissipation.max_abs_speed, ")")
	end
	"""
	flux_lax_friedrichs
	See [`FluxLaxFriedrichs`](@ref).
	"""
	const flux_lax_friedrichs = FluxLaxFriedrichs()

is computed

Trixi.jl/src/equations/compressible_euler_2d.jl

Line 1413 in e93de2a

λ_max = max(abs(v_ll), abs(v_rr)) + max(c_ll, c_rr)

compared to the way employed in the StepsizeCallback:

Trixi.jl/src/callbacks_step/stepsize_dg2d.jl

Line 18 in e93de2a

lambda1, lambda2 = max_abs_speeds(u_node, equations)

which uses the actual (sharp) true eigenvalues of the flux Jacobian:

Trixi.jl/src/equations/compressible_euler_2d.jl

Lines 1863 to 1868 in e93de2a

	@inline function max_abs_speeds(u, equations::CompressibleEulerEquations2D)
	rho, v1, v2, p = cons2prim(u, equations)
	c = sqrt(equations.gamma * p / rho)
	return abs(v1) + c, abs(v2) + c
	end