Redesign signatures of compilation targets

[jakobj]

This PR aims at making the usage of the results from various to_... functions more intuitive. It also improves their documentation. In the following the new usage is described in more detail.

to_func -> f:

• f expects the same number of arguments as the graph has input
• each argument must be a single value
• f returns a list of values, one for each output
• if only one output is present it'll return only it's output value; not a list with one entry
  example usage:

f = ind.to_func()
y = f(1.0, 2.0)

to_numpy -> f:

• f expects the same number of arguments as the graph has input
• each argument must be a single numpy array of identical lengths
• f returns a list of arrays, one for each output; each array will have the same length as all input arrays
• if only one output is present it'll return only it's output array; not a list with one array
  example usage:

f = ind.to_numpy()
y = f(np.array([1.0, 1.1]), np.array([2.0, 2.1]))

to_sympy -> List[sympy expr] or single sympy expr

• depending on the number of outputs, either a list of sympy expressions is returned or, for a single output, a single sympy expression

closes #287

[coveralls]

Coverage decreased (-0.05%) to 94.332% when pulling 13c8362 on jakobj:enh/redesign-call-sig into 3411cba on Happy-Algorithms-League:master.

[mschmidt87]

I think this branch needs to be rebased on master to pull in the changes in the continuous integration and make the checks run.

[jakobj]

I think this branch needs to be rebased on master to pull in the changes in the continuous integration and make the checks run.

i rebased, thanks! ready for review :)

[jakobj]

ping @mschmidt87 @HenrikMettler

[HenrikMettler]

Just two minor formulations which might be improved

[mschmidt87]

Looks good in principle, thanks for doing this, looks like a tedious job. Added some suggestions.

[jakobj]

thanks @mschmidt87 @HenrikMettler for your comments, i've included most of them. please have another look.

[jakobj]

ok, simplified the code in question @mschmidt87

also squashed all commits, let's see what CI thinks