Make package installation automatic and deprecate manual installation

[LilithHafner]

This has three benefits

• Less code (if we exclude deprecated code)
• Easier setup (there is no install step for de or any of the GPU backends)
• Don't have to install and precompile DifferentialEquations to use from diffeqpy import ode

The principle is to use

try
    import Package
catch
    import Pkg
    Pkg.add("Package")
    import Package
end

---

Before

>>> from juliacall import ode
...
juliacall.JuliaError: ArgumentError: Package OrdinaryDiffEq not found in current path.
- Run `import Pkg; Pkg.add("OrdinaryDiffEq")` to install the OrdinaryDiffEq package. # bad hint
...
>>> import juliacall
>>> juliacall.install() # Installs de and ode
...
>>> from juliacall import ode

After

>>> from juliacall import ode # Installs de

---

Also changes the API of the internal load_julia_packages to be more reasonable (takes a list of strings instead of a string with commas in it) and more robust ("just works" regardless of instillation status)

TODO: support manual updates with diffeqpy.update() => Pkg.up() or similar.