[parsing] Slightly improve joint parsing diagnostics

[jwnimmer-tri]

Mostly, just clarify our plans for future work, and be sure to notice when SDFormat adds new joint types (via C++ switch-enum warnings).

To provide more context for the TODOs -- as discussed in slack, to the extent that we can best-effort deal with unimplemented features (while warning about it to the user), that is probably the best UX. It will allow the user to actually create a MbP and play with it, resolving the warnings later in due course.

Closes #16784.

---

This change is 

[jwnimmer-tri]

+@sammy-tri for feature review, please.

[sammy-tri]

+@xuchenhan-tri for platform review by rotation (Monday)

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: LGTM missing from assignee xuchenhan-tri(platform) (waiting on @xuchenhan-tri)

[xuchenhan-tri]

Reviewed 1 of 1 files at r1, all commit messages.
Reviewable status: 1 unresolved discussion, LGTM missing from assignee xuchenhan-tri(platform) (waiting on @jwnimmer-tri)

---

-- commits line 5 at r1:
BTW, why is switch enum warning preferred to the default throw approach used before?

---

multibody/parsing/detail_sdf_parser.cc line 621 at r1 (raw file):

    case sdf::JointType::CONTINUOUS: {
      // TODO(#14747) Use an unlimited multibody::RevoluteJoint here.
      diagnostic.Error(fmt::format(

nit, are these covered by unit tests?

[jwnimmer-tri]

Reviewable status: LGTM missing from assignee xuchenhan-tri(platform) (waiting on @sammy-tri and @xuchenhan-tri)

---

-- commits line 5 at r1:

Previously, xuchenhan-tri wrote…

BTW, why is switch enum warning preferred to the default throw approach used before?

When code uses a switch() statement over an enum and the switch does not contain a default: case, then the compiler will emit a warning about any enum values that have not been mentioned as cases. In Drake's build settings, we promote that warning to an error.

This means that is upstream libsdformat ever adds a new enumeration value, Drake developers will be forced to explicitly consider what we should do about it, because CI will bark during the libsdformat pin upgrade.

---

multibody/parsing/detail_sdf_parser.cc line 621 at r1 (raw file):

Previously, xuchenhan-tri wrote…

nit, are these covered by unit tests?

Done.

[xuchenhan-tri]

Reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: complete! all discussions resolved, LGTM from assignees sammy-tri(platform),xuchenhan-tri(platform) (waiting on @jwnimmer-tri)

---

-- commits line 5 at r1:

Previously, jwnimmer-tri (Jeremy Nimmer) wrote…

When code uses a switch() statement over an enum and the switch does not contain a default: case, then the compiler will emit a warning about any enum values that have not been mentioned as cases. In Drake's build settings, we promote that warning to an error.

This means that is upstream libsdformat ever adds a new enumeration value, Drake developers will be forced to explicitly consider what we should do about it, because CI will bark during the libsdformat pin upgrade.

Nice, thanks for the explanation