High-excitation code paths in MSD are untested

**Describe the bug**
The work in #3900 highlights that the "general" code path for excitations is untested.

**To Reproduce**
Run the tests.

**Expected behavior**
Unit or deterministic test hits the code path.

**Additional context**

An easy solution would be to add an MSD only integration test (short and deterministic) using FCI on a pseudoatom. Better would be properly unit testing the general code paths.

In view of the offload work and still-to-be-proven nature of offload compilers, these code paths should be tested.