Sunsetting esa-ptr ?

MR !12 provide a first attempt to include a local PTR solver using the PTWrapper package. Originally the esa-ptr package was supposed to use the local AGM solver only when it was available and fallback to the online API when the solver was not available (because the solver was not supposed to be publicly distributed). This approach is no longer necessary now that PTWrapper is available on pypi and the need for the esa-ptr package as a middle-man is not obvious and may no longer be required.

Here is a list of the pros/cons I can think for esa-ptr vs. PTWrapper:

Pros:

• Simpler interface to submit a PTR (agm_simulation('JUICE CREMA 5.0', 'example.ptx', 'JUICE_API'))
• Internal validation before the submission (to catch errors before the solver)
• HTML representation of the results in jupyter notebooks
• OS independent (PTWrapper does not work on Windows)
• Online documentation
• Stronger CI and test coverage

Cons:

• Duplicated layer that add complexity (one more version to track)
• Slower updates because it's a 2 steps process
• More difficult to keep track of the default parameters
• May be inconsistent between the local and the online solver
• I/O limitation of the API that don't resolve complex PTR
• Very demanding on the Pointing Tool
• If PTWrapper is not available, it requires an internet connection the pointing tool

Let me know what is your feeling on this to see if it is relevant to sunset this package in favor of PTWrapper?

cc: @mcosta, @nmanaud @lpenasa