Filter out subcomponents in build_components_refresh

[jwnimmer-tri]

Closes #6464.

Ideally we would continue to work on components still marked as # unpackaged in the list, but for now at least it seems worth locking in this stage before moving on to fix the stragglers.

---

This change is 

[jwnimmer-tri]

+@jamiesnape for feature review, please.
+@ggould-tri for platform review per schedule, please.

[ggould-tri]

, notes below.

---

Reviewed 3 of 3 files at r1.
Review status: all files reviewed at latest revision, all discussions resolved.

---

tools/install/libdrake/build_components.bzl, line 8 at r1 (raw file):

#
# When adding new components to the package, please also add the licenses for
# any new external dependencies to :external_licenses.

BTW -- Author's discretion as it is not part of this PR, but it seems odd to me to put "when adding new components" right after "don't update this list". Perhaps this comment is in the wrong place?

---

tools/install/libdrake/build_components_refresh.py, line 123 at r1 (raw file):

                one_label, component_labels)
            if sans_colon_label:
                line = '    "{}",'.format(sans_colon_label)

This is going to disturb the sort you do above (line 64) , resulting in surprising sorts like the position of //multibody/rigid_body_plant in the generated file. I'm a little bit surprised that buildifier allowed it.

Perhaps it would be better to sort after this step rather than in _find_libdrake_components?

---

Comments from Reviewable

[jamiesnape]

---

Reviewed 3 of 3 files at r1.
Review status: all files reviewed at latest revision, 2 unresolved discussions.

---

Comments from Reviewable

[jwnimmer-tri]

Review status: 1 of 3 files reviewed at latest revision, 2 unresolved discussions.

---

tools/install/libdrake/build_components.bzl, line 8 at r1 (raw file):

Previously, ggould-tri wrote…

BTW -- Author's discretion as it is not part of this PR, but it seems odd to me to put "when adding new components" right after "don't update this list". Perhaps this comment is in the wrong place?

Done. Wow, that comment was old and very, very stale.

---

tools/install/libdrake/build_components_refresh.py, line 123 at r1 (raw file):

Previously, ggould-tri wrote…

This is going to disturb the sort you do above (line 64) , resulting in surprising sorts like the position of //multibody/rigid_body_plant in the generated file. I'm a little bit surprised that buildifier allowed it.

Perhaps it would be better to sort after this step rather than in _find_libdrake_components?

Done. Great point. It even fixed a bug (rigid_body_plant should have been commented as unpackaged).

---

Comments from Reviewable

[jamiesnape]

Reviewed 2 of 2 files at r2.
Review status: all files reviewed at latest revision, 1 unresolved discussion.

---

Comments from Reviewable

[jwnimmer-tri]

I tested this locally; the only difference in the install is that include/drake/multibody/multibody_tree:/parsing/*.h are newly installed, since that package is now mentioned due to the refresh. (The PRs that added that code to master did not run build_components_refresh.)