SIEM documentation + PETSC compilation updates

[williameaton]

The following was added/edited:

1. Documentation added for full-gravity simulations, in particular for the simulation of PEGS. This seemed to fit most naturally in 'gravity calculations' but could be added elsewhere instead.
2. For PEGS computations we want to avoid a Gaussian STF since it causes energy to arrive too early before the P wave. Added a sine-squared STF functionality as part of the Par_file. Perhaps you would prefer this to be a case of using the external STF function instead of a flag specifically for this STF type.
3. Outputs elastic acceleration seismograms when full-gravity switch is on (needed for PEGS calculations)
4. Updated some PETSC features since it wasn't compiling, in particular:
   • Some of the calls such as VecGetArrayF90 seem to have been re-refactored back to VecGetArray so have updated the subroutine names
   • MatSetValues seems to be stricter on requiring a 1D array (previously 2D) so inputting flattened array, and also parsing single values as length 1 arrays
   • MatMPIAIJSetPreallocation being stricter on needing NULL_ARRAY types instead of NULL.
   • Perhaps some of these PETSC issues are compiler dependent, but this is compiling with the most recent distribution of PETSC

[danielpeter]

great, thanks for this gravity update!

the only change I can see is in the documentation. I would use now "SPECFEM3D Globe" instead of "SPECFEM" and "SPECFEM3D_GLOBE".