Pyscf UCAS wavefunctions unsupported by converter (partial fix suggested)

**Describe the bug**
Pyscf UCAS wavefunctions are not supported by the PyscfToQmcpack converter.

**To Reproduce**
call `PyscfToQmcpack.savetoqmcpack` with a UCAS object (UCASCI or UCASSCF)

**Expected behavior**
generate hdf5 orbs and multidet file for qmcpack

I discussed this with @jtkrogel and @anbenali on Slack, but I wanted to mention it here in case that disappears beyond the 90-day time horizon.
There've been similar instances where we've found a temporary working solution to a problem and then needed to solve the same thing again several months/years later.

Here is a tarball containing some example inputs from @jtkrogel that I modified (along with using a modified converter) to produce hdf5 orbs and multidets for QMCPACK (also included here): [nitrogen_uhf_ucas.tar.gz](https://github.com/user-attachments/files/21691228/nitrogen_uhf_ucas.tar.gz)

Here is a diff of some of the needed changes in the converter: [PyscfToQmcpack_UCAS.patch](https://github.com/user-attachments/files/21690680/PyscfToQmcpack_UCAS_patch.txt)

Summary of changes:
- detect `UCASCI` and `UCASSCF` as unrestricted
- detect `UCASCI` and `UCASSCF` as multidet
- modify the way multidet objects are handled to more closely mirror the way we handle `Restricted` and `PBC`
- correctly handle `nstate` in unrestricted multidet h5 generation (also see note below about this)

Remaining issues:
- the converter still expects the CAS object to have `mo_energy`. This isn't present in UCAS objects.
- We should handle this in the converter, but a hacky workaround is to assign it before calling the converter.
Something like:
```
mycas.mo_energy = ucas_fock_diag(mycas)
```
with this function to compute diagonal elements of the "fock matrix" (not sure if this is even correct; it's just a quick fix, so use at your own risk)
```
def ucas_fock_diag(mc):
rdms=mc.make_rdm1s()
vj,vk=mc._scf.get_jk(mc.mol,rdms)
veff=vj[0]+vj[1]-vk
fock = mc.get_hcore() + veff
c = np.array(mc.mo_coeff) # can use mc.mo_coeff for UCASCI, but for UCASSCF it's a tuple, so convert to array
fock_mo_diag = np.einsum('sui,suv,svj->si',c.conj(),fock,c)
return fock_mo_diag
```
- somewhat related to this is the way we assign `MultiDet/nstate` in the generated hdf5 file. It looks like it's being set as the number of AOs [here](https://github.com/QMCPACK/qmcpack/blob/7f1dd802c7d5c003c79779c3529676a04d7539df/src/QMCTools/PyscfToQmcpack.py#L623), but I would have thought that this would be the number of MOs (i.e. the space corresponding to the occupation bitstrings). It might be fine already, but I noticed it while applying the UCAS fixes and thought it was worth checking