[fem] Refactor ForceDensityField

[xuchenhan-tri]

The goal of the commit is the remove the dependency of the fem library from the tree library so that later we create deformable multibody element (in the tree library) that depends on fem.

In particular, add a ForceDensityFieldImpl class so that the interface ForceDensityField no longer depends on tree and can be left in the fem library.

Towards #23029

---

This change is 

[xuchenhan-tri]

+@sherm1 for feature review please.

Reviewable status: LGTM missing from assignee sherm1(platform), needs at least two assigned reviewers

[sherm1]

BTW lint is unhappy with point_source_force_field.h, causing CI failures.

First pass done. A couple of thoughts:

• It seems odd to have the Impl class derived from the nice-named class. Consider renaming like this:
  ForceDensityField -> ForceDensityFieldBase
  ForceDensityFieldImpl -> ForceDensityField
  Then concrete implementations can continue to derive from ForceDensityField.
• I can't tell if there is a problem with Clone() now. There are data members in both the base and impl classes, but the impl class doesn't do anything about cloning. Presumably this is left to the concrete classes. Are there unit tests somewhere that can show that those DoClone()s actually work properly with the new intermediate class stuck in there?

Reviewed 18 of 18 files at r1, all commit messages.
Reviewable status: 1 unresolved discussion, LGTM missing from assignee sherm1(platform), needs at least two assigned reviewers

---

multibody/tree/force_density_field_impl.h line 112 at r1 (raw file):

  const internal::MultibodyTreeSystem<T>* tree_system_{nullptr};
  ForceDensityType density_type_{ForceDensityType::kPerCurrentVolume};

BTW this field density_type_ is also present in the now-base-class ForceDensityField. Is that intentional?

[xuchenhan-tri]

I added a new unit test file with some new tests in r2. I'll push the name change in r3.

Reviewable status: 1 unresolved discussion, LGTM missing from assignee sherm1(platform), needs at least two assigned reviewers, commits need curation (https://drake.mit.edu/reviewable.html#curated-commits) (waiting on @xuchenhan-tri)

---

multibody/tree/force_density_field_impl.h line 112 at r1 (raw file):

Previously, sherm1 (Michael Sherman) wrote…

BTW this field density_type_ is also present in the now-base-class ForceDensityField. Is that intentional?

No! Removed, thanks!

[xuchenhan-tri]

I pushed the name change in r3, but I'm not a huge fan of the new names. The name exposed in public APIs is now ForceDensityFieldBase (e.g. in DeformableModel::AddExternalForce)

Reviewable status: 1 unresolved discussion, LGTM missing from assignee sherm1(platform), needs at least two assigned reviewers, commits need curation (https://drake.mit.edu/reviewable.html#curated-commits) (waiting on @xuchenhan-tri)

[sherm1]

ForceDensityField also has public APIs so it's not obvious to me that there's a perfect solution. I do like how this looks in your test case:

class ConstantForceDensityField final : public ForceDensityField<double>

Anyway, thanks.

+@ggould-tri for platform review per rotation, please

Reviewed 22 of 22 files at r2, all commit messages.
Reviewable status: LGTM missing from assignee ggould-tri(platform), commits need curation (https://drake.mit.edu/reviewable.html#curated-commits) (waiting on @xuchenhan-tri)

[xuchenhan-tri]

@drake-jenkins-bot mac-arm-sonoma-clang-bazel-experimental-release please.

[ggould-tri]

Reviewed 1 of 18 files at r1, 22 of 22 files at r2, all commit messages.
Reviewable status: 2 unresolved discussions, commits need curation (https://drake.mit.edu/reviewable.html#curated-commits) (waiting on @xuchenhan-tri)

---

multibody/plant/deformable_model.cc line 588 at r2 (raw file):

       force_densities_) {
    /* We know that the static cast is safe because all concrete force density
     field is derived from ForceDensityField. */

typo: Grammar

Suggestion:

    /* We know that the static cast is safe because all concrete force density
     fields are derived from ForceDensityField. */

---

multibody/fem/force_density_field_base.h line 38 at r2 (raw file):

class ForceDensityFieldBase {
 public:
  virtual ~ForceDensityFieldBase() = 0;

minor: I'm not sure what you are trying to accomplish by pure-virtualing this in the header and then setting it default in the cc file?

[ggould-tri]

Reviewed 1 of 1 files at r3, all commit messages.
Reviewable status: 1 unresolved discussion

[xuchenhan-tri]

Reviewable status: 1 unresolved discussion

---

multibody/plant/deformable_model.cc line 588 at r2 (raw file):

Previously, ggould-tri wrote…

typo: Grammar

Done

---

multibody/fem/force_density_field_base.h line 38 at r2 (raw file):

Previously, ggould-tri wrote…

minor: I'm not sure what you are trying to accomplish by pure-virtualing this in the header and then setting it default in the cc file?

Pure virtual to make this class abstract and setting its implementation as default because there's nothing special needed for the destructor. To be clear, a definition in the cc file is still required for a pure virtual desctructor.