Intern frequently used PyObject?

While examing the `gc` of my code I saw a LOT of PyObject being created and garbage collected. (I modified `PyCall` so as to print the object at finalizing)

I am wondering if `PyCall` can automatically cache many frequently used objects, such as

```
PyObject([])
PyObject(None)
PyObject(True)
PyObject(False)
```

and maybe even `PyObject(0), ... PyObject(255)`.

For example, I have a [Julia wrapper](https://github.com/colinfang/PyLogging.jl) for python `logging`. Each logging would generate 7 `PyObjects`. That's a bit too many.

```
julia> PyLogging.basicConfig()

julia> @warning logger "ll"
WARNING:root:ll

julia> gc()
"Start pydecref PyObject (30,)"
"Start pydecref PyObject True"
"Start pydecref PyObject 30"
"Start pydecref PyObject 'll'"
"Start pydecref PyObject []"
"Start pydecref PyObject (30, 'll', [])"
"Start pydecref PyObject None"
```

Start by inspecting PyCall's PyObject construction and finalization paths, especially the constructors for None, booleans, empty lists, tuples, and small integers. Determine how frequently created objects could be cached without breaking reference management, then verify that repeated logging calls create fewer temporary PyObjects and still finalize safely.

Rédigé par le modèle d'indexation à partir du texte de l'issue.