More adaptations for Taal/Taam

[ranocha]

julia> begin
           t_end = 10.0
           setup = "euler_sedov_blast_wave_shockcapturing_amr"
           taam = Trixi.run("examples/2d/parameters_" * setup * ".toml", t_end=t_end)
           trixi_include("examples/2d/elixir_" * setup * ".jl", tspan=(0.0, t_end))
           taal = analysis_callback(sol)
           display(taal.l2 - taam.l2)
           display(taal.linf - taam.linf)
       end
...
4-element StaticArrays.SArray{Tuple{4},Float64,1,4} with indices SOneTo(4):
 1.176836406102666e-14
 2.4980018054066022e-15
 8.326672684688674e-17
 7.771561172376096e-16
4-element StaticArrays.SArray{Tuple{4},Float64,1,4} with indices SOneTo(4):
 -3.019806626980426e-14
  4.8183679268731794e-14
 -7.327471962526033e-15
  3.552713678800501e-15

Seems to be okay after 1446 time steps.

This example is very sensitive to many parameters and will likely crash if you change anything, e.g. some fluxes etc. Hence, I think some differences occurring for bigger CFL numbers might really be influenced by floating point errors etc.