Buggy hybrid initialization in FGRAYRAD

### What happened?

When trying to initialize a hybrid simulation with the FGRAYRAD compset, instead of initializing from the run specified in RUN_REFCASE etc, it initializes again from the analytic initial condition. A branch run seems to work OK. This has been instigated by this forum post: https://bb.cgd.ucar.edu/cesm/threads/hybrid-run-in-gray-radiation-aquaplanet.10956/

### What are the steps to reproduce the bug?

Run one year of FGRAYRAD. Start again with a hybrid initialization from the first run and if you output daily average T and look at it, you'll see that day one of year 2 goes back to being symmetric in longitude, without any weather systems because it's starting back again from the zonally symmetric analytic initial condition.

### What CAM tag were you using?

cesm2_3_alpha17f

### What machine were you running CAM on?

CISL machine (e.g. cheyenne)

### What compiler were you using?

Other (please specify below)

### Path to a case directory, if applicable

/glade/work/islas/cesm2_3_alpha17f/runs/FGRAYRAD.001

### Will you be addressing this bug yourself?

No

### Extra info

_No response_