Neumann instead of Dirichlet boundary for 2nd order elastic wave free surface condition

[sframba]

The free-surface condition for the elastic wave equation is the Neumann one, i.e.,

$\left.\frac{\partial u{x,y,z}}{\partial n}\right|{\partial\Omega}=0$

Currently, the Dirichlet condition is applied instead:

$\left. u{x,y,z}\right|{\partial\Omega}=0$

leading to an unphysical polarity change in the reflected wave. To correct this, one can simply remove the forced zeroing-out of boundary node displacements.

[codecov[bot]]

Codecov Report

All modified and coverable lines are covered by tests :white_check_mark:

Project coverage is 53.21%. Comparing base (b92d789) to head (5265150).

Additional details and impacted files

```diff @@ Coverage Diff @@ ## develop #3072 +/- ## =========================================== - Coverage 53.21% 53.21% -0.01% =========================================== Files 989 989 Lines 83604 83602 -2 =========================================== - Hits 44494 44492 -2 Misses 39110 39110 ```

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[sframba]

We are still having trouble accessing LC -- can someone please do the rebaseline if this gets to the top of the merge queue? Only the elastic wave solver should be affected. Thanks!