Archive rest directory gets screwed up when rerunning a simulation that already stored at least two run segments

### Brief summary of bug

I found this from working with @katierocci at the parallel work session. She ran two segments of a simulations saving restarts at years 201 and 401. Then she realized she needed to rerun the simulation with a new parameter file. She left the previously generated archive for the case with a rest directory for both year 201 and for 401.

The rerun from 0 to year 201 went fine, and the files at the end where fine. But, the st_archive messed up the archive/rest/401 directory and didn't put the new rerun files into the archive/rest/201 directory.

### General bug information

**CTSM version you are using:** ctsm5.2.028

**Does this bug cause significantly incorrect results in the model's science?** No

**Configurations affected:** Rerunning a simulation with output already saved

### Details of bug

After the rerun of year 201, it put rpointer files for year 201 into the archive/rest/401 directory and left the 401 restart files there. And the year 201 restart files seemed to have been deleted, as they weren't in the run directory anymore nor copied over in the archive/rest/201 directory.

If this glitch is addressed it would need to be updated in cime. Really all that might need to be done is for the st_archive to fail if it finds the restart files with the wrong or inconsistent dates. Or for the simulation to check for this at preview_namelist time?

The workaround for this is to delete the 2nd and later segment archive/rest directories.