pydrake: Expose pybind11 version information

[EricCousineau-TRI]

Addresses part (b) of #7856, supersedes #8107.

---

This change is 

[EricCousineau-TRI]

+@jamiesnape for feature review, please.

---

Review status: 0 of 4 files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable

[jwnimmer-tri]

The new write_file.bzl and the pre-existing tools/workspace/generate_file.bzl probably need some de-duplication.

[EricCousineau-TRI]

Done. Keeping generate_file where it is for now, can move it later if need be.

---

Review status: 0 of 7 files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable

[jamiesnape]

---

Reviewed 8 of 8 files at r1.
Review status: all files reviewed at latest revision, 1 unresolved discussion.

---

bindings/pydrake/util/test/pybind11_version_test.py, line 7 at r1 (raw file):

from __future__ import print_function

import pydrake.util.pybind11_version as mut

BTW Not a great fan of mut, the meaning is not obvious without some thought, but I guess it is convention here.

---

Comments from Reviewable

[EricCousineau-TRI]

+@sammy-tri for platform review, please.

---

Review status: all files reviewed at latest revision, 1 unresolved discussion.

---

bindings/pydrake/util/test/pybind11_version_test.py, line 7 at r1 (raw file):

Previously, jamiesnape (Jamie Snape) wrote…

BTW Not a great fan of mut, the meaning is not obvious without some thought, but I guess it is convention here.

OK I'd prefer to keep it for now, but will take note. Thanks!

---

Comments from Reviewable

[sammy-tri]

---

Reviewed 8 of 8 files at r1.
Review status: all files reviewed at latest revision, all discussions resolved, some commit checks failed.

---

Comments from Reviewable

[EricCousineau-TRI]

Review status: 4 of 8 files reviewed at latest revision, all discussions resolved, some commit checks pending.

---

bindings/pydrake/util/BUILD.bazel, line 146 at r2 (raw file):

    name = "pybind11_version_py",
    srcs = ["pybind11_version.py"],
    # `ctx.actions.write` will always produce an exectuable file, regardless of

FYI @jwnimmer-tri Can I ask what you'd like to do here? Just always nolint any generated py files?
I checked genfiles, and all generated files are executable.

---

Comments from Reviewable

[EricCousineau-TRI]

Review status: 4 of 8 files reviewed at latest revision, 1 unresolved discussion.

---

bindings/pydrake/util/BUILD.bazel, line 146 at r2 (raw file):

Previously, EricCousineau-TRI (Eric Cousineau) wrote…

FYI @jwnimmer-tri Can I ask what you'd like to do here? Just always nolint any generated py files?
I checked genfiles, and all generated files are executable.

Er, that was grammatically unclear: "Should we always add tags = ["nolint"] for any generated py files?"

---

Comments from Reviewable

[jwnimmer-tri]

bindings/pydrake/util/BUILD.bazel, line 146 at r2 (raw file):

Previously, EricCousineau-TRI (Eric Cousineau) wrote…

Er, that was grammatically unclear: "Should we always add tags = ["nolint"] for any generated py files?"

OK Yeah, genfiles are always executable for some reason. Adding nolint is fine. (The longer term game, which also applies to cxx files, is to teach the linter that things in genfules shouldn't have the shebang check applied.)

---

Comments from Reviewable