Actually initialize zero_acceleration_kinematics_placeholder_ to zero

[Brian-Acosta]

This comment in multibody_plant.h is lying about zero_acceleration_kinematics_placeholder_ getting populated with zeros :

  // When use_sampled_output_ports_ is true, then during Finalize() we populate
  // this with all-zero data, for use when the State has not yet been stepped.

This PR fixes this issue to avoid sending NaNs from the body_spatial_accelerations output port on the first step of a discret-time MBP simulation.

---

This change is 

[sherm1]

@drake-jenkins-bot ok to test

[sherm1]

+@jwnimmer-tri for a quick (4 line) sampled-output-ports-related feature review, please. Is this a one-off or are there other instances we should fix?

Reviewable status: LGTM missing from assignee jwnimmer-tri(platform), needs at least two assigned reviewers, missing label for release notes (waiting on @Brian-Acosta)

[jwnimmer-tri]

I am happy to review. But it won't be as quick as four lines -- I will also need to develop and push the missing unit test.

[jwnimmer-tri]

+@sherm1 for feature review, please.

(We'll assign a platform reviewer who is not me, later on.)

Reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: LGTM missing from assignees jwnimmer-tri(platform),sherm1(platform), commits need curation (https://drake.mit.edu/reviewable.html#curated-commits) (waiting on @Brian-Acosta)

[sherm1]

Feature

Reviewed 2 of 2 files at r2, all commit messages.
Reviewable status: LGTM missing from assignee jwnimmer-tri(platform), commits need curation (https://drake.mit.edu/reviewable.html#curated-commits) (waiting on @Brian-Acosta)

[jwnimmer-tri]

+@ggould-tri for platform review per schedule (Tuesday), please.