Restart from checkpoint fails when some weights have been equilibrated, but not all

[ajfriedman22]

Describe the bug
When you attempt to restart a MT-REXEE or EEXE simulation from a checkpoint file when some of the simulations have equilibrated and others have not an error occurs. The text of the error is below. I have discovered that the error occurs because the simulation which equilibrated before the checkpoint is counted as a fixed-weight simulation rather than as a variable-weight simulation which has already equilibrated.

** Error Text **

An error occurred on rank 0:
Traceback (most recent call last):
File "/projects/anfr8476/code/ensemble_md/ensemble_md/cli/run_REXEE.py", line 246, in main
_ = REXEE.combine_weights(weights, print_values=False)[1] # just to print the combiend weights # noqa: E501
File "/projects/anfr8476/code/ensemble_md/ensemble_md/replica_exchange_EE.py", line 1353, in combine_weights
weights_modified[i] = self.equilibrated_weights[i]
ValueError: could not broadcast input array from shape (0,) into shape (9,)

---

To Reproduce

1. Start a variable weight simulation
2. When at least one simulation has equilibrated but not all of them stop the simulation (A checkpoint file must be saved after at least one simulation is equilibrated)
3. Restart the simulation from the checkpoint

Proposed Fix
We can add an additional checkpoint .npy file which will save the equilibration times if individual simulations have already equilibrated and then load these in when we restart from checkpoint.