operators: Explicitly expose py::hash(py::self)

[EricCousineau-TRI]

Add warnings about extending STL

When using pybind11 for Drake, I found that I had to write:

cls.def(py::detail::hash(py::self));

instead of what the unittest wrote:

pybind11/tests/test_operator_overloading.cpp

Line 110 in a54eab9

.def(hash(py::self))

This explicitly exposes the overload in the pybind11 namespace.
Additionally, I was mislead into thinking that you may be able to overload std::abs. Per these docs, it does not appear to be valid (nor could I get it to work straight off the bat):
https://en.cppreference.com/w/cpp/language/extending_std

Follow-up to #1034

[rwgk]

Looks great to me!
If easy: may be nice to undo whitespace formatting change after line 48 (old) of test_operator_overloading.py, so that it is obvious that the values aren't changed by this PR. (I'd also remove the extra empty line at the end.)

[EricCousineau-TRI]

@rwgk Thanks! However, it was still a tad hard for me to "add" in the new value b/c of whitespace changes.
Would it be OK if I instead curated my commits?
e.g. make minimal set of whitespace changes, no functional changes, and then add the functional change on top of that? Then bisecting / git history would be preserved!

[rwgk]

Only what you feel is easy and reasonable. It's nice to minimize distracting whitespace diffs but only if the effort to achieve that is reasonable.

…

On Thu, May 21, 2020 at 12:35 PM Eric Cousineau ***@***.***> wrote: @rwgk <https://github.com/rwgk> Thanks! However, it was still a tad hard for me to "add" in the new value b/c of whitespace changes. Would it be OK if I instead curated my commits? e.g. make minimal set of whitespace changes, no functional changes, and then add the functional change on top of that? Then bisecting / git history would be preserved! — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#2217 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAFUZADYFBIYETHMP2GMUATRSV67XANCNFSM4NEMV7PQ> .

[EricCousineau-TRI]

Took about 3min to split the commits apart (and make sure each one tested correctly).
Lemme know what you think!

[rwgk]

Thanks Eric!

[wjakob]

Looks great, thanks @EricCousineau-TRI and @rwgk for the review!