Develop a performance profiling suite

### Are there any linked Issues or Pull Requests?

#80, #104

### Brief description

To monitor performance, a profiling suite should be created. This would likely have the following characteristics:

- Run on a weekly basis
- Include a C192, C224, and C896 jobs up to 108 nodes.
- C192 with no IO
- C224 to include IO but also have no IO version
- Include jobs to monitor XIOS server performance
- With and without diagnostics
- Include measures of timestepping time _without_ XIOS
- Test `production` level optimisations
- Use Vernier (#80)
- Use CrayPat (#104)
- Process the output from Vernier, CrayPat, etc., to display relevant information

The suite _may_ include additional features such as:

- `MPICH_MPIIO_STATS` output
- `MPICH_MPIIO_TIMERS` output

### Further details of the issue.

This issue has several distinct facets:

- [ ] The creation of tests will require input from several teams including TCD and HPC Opt
- [x] ~~The inclusion of tests in the weekly suite must be approved by SSD~~
- The suite will be standalone:
- jobs will be too large for rose-stem
- rose-stem does not need additional options added
- [x] Vernier changes must be merged
- [ ] CrayPat needs to be fixed