Unit conversion inaccuracies; convert everything to atomic units

[jagot]

Due to the unit conversions that go on under the hood, we get issues such as the following:

julia> @field(F) do
           I₀ = 1.0
           T = 1.0
           τ = 8.0
           σmax = 6.0
       end
Linearly polarized field with
  - I₀ = 1.0000e+00 au = 3.5094452e16 W cm^-2 =>
    - E₀ = 1.0000e+00 au = 514.2207 GV m^-1
    - A₀ = 0.1592 au
  – a Fixed carrier @ λ = 7.2516 nm (T = 24.1888 as, ω = 6.2832 Ha = 170.9742 eV, f = 41.3414 PHz)
  – and a Gaussian envelope of duration 193.5107 as (intensity FWHM; ±6.18σ)
  – and a bandwidth of 0.3466 Ha = 9.4307 eV ⟺ 2.2803 PHz ⟺ 7.5588 Bohr = 399.9921 pm
  – Uₚ = 0.0063 Ha = 172.3181 meV => α = 0.0253 Bohr = 1.3404 pm

julia> F.I₀
1.0

julia> F.E₀
0.9999999540483009

The last should obviously be exactly unity.

The problem arises because I₀ is first converted into SI units, then E₀ is computed using

ElectricFields.jl/src/field_dsl.jl

Line 54 in 39a7586

E₀ = √(2I₀/(u"ε0"*u"c"))

and then subsequently converted back to atomic units. This is silly, and the detour via SI units should be avoided.