Components of `calc_Obhukov_length` are most of the cost of `simulate_microclimate`

[rafaqz]

Mostly these logs and powers:

Microclimate.jl/src/boundary_layer.jl

Line 380 in d1efb50

return 2.0 * log((1.0 + x) / 2.0) + log((1.0 + x^2) / 2.0) - 2.0 * atan(x) + π / 2.0

Microclimate.jl/src/boundary_layer.jl

Line 307 in d1efb50

return 0.62 / (ustrip(u"cm", z0) * ustrip(u"cm/minute", u_star) / 12)^(9//20)

We could look into if any of these are not affected by loop variables and precompute them, and similar.

[rafaqz]

With type stability fixes in #82 the profile is looking pretty good, but now calc_Objukov_length is clearly about 70% of the total run! Each large green/blue block in that row where its named is just an other call to it from a different location.