Converting GAMESS output with multistate calculation

[neelravi]

This pull request offers a solution (issue #39) to handle the conversion of gamess or gaussian multistate calculation into multiple trexio files.

The summary of the work:

• The original code is put under the loop over nstates for s in range(res.num_states):
• Trexio files are generated for each state with names ending with _state_0.hdf5 or _state_1.hdf5
• Tests are modified for these new naming convention
• The following new block is added to store the state specific information:

        # State group
        # ---------
        trexio.write_state_num(trexio_file,res.num_states)
        trexio.write_state_current_label(trexio_file, "State "+ str(s) + " of " + str(res.num_states-1))
        trexio.write_state_label(trexio_file, [f"state{i}" for i in range(res.num_states)])
        trexio.write_state_file_name(trexio_file, [f"{trexio_basename}_state_{i}.hdf5" for i in range(res.num_states)])

        # CSF group
        # ---------
        try:
            num_csfs = len(res.csf_coefficients[s])
        except:
            num_csfs = len(res.det_coefficients[0])

        offset_file = 0
        trexio.write_csf_coefficient(trexio_file, offset_file, num_csfs, res.csf_coefficients[s])

[q-posev]

Thanks @neelravi !

I left a few comments. My main concern is about custom _state suffix which breaks the way the converter used to work. Not sure if this is the way to go.

Also, if this PR closes issue #39 , you can add "- Closes #39" line to the PR description and this issue will be closed automatically when the PR is merged.

[q-posev]

I don't understand why the -b hdf5 has to be removed here. The test should work as it used to.

[q-posev]

I don't think it is a good idea to do it this way. The user provided trexio_filename via CLI for a reason - they would expect it to be the name of the output TREXIO file. The moment you start to modify this logic with custom suffixes (like _state used here) - it becomes confusing and requires extra effort from the user. Now the ground state trexio files are all suffixed with _state_0 which basically breaks how the converter used to work and might affect the user workflows.
I personally think that it might be better to provide the state ID via CLI and let the user do the the conversion N times where N is the desired number of states. Yes, it will require N calls to resultsFile but at least the user will have full control of the pipeline. @scemama @neelravi what do you think?

[q-posev]

Are you sure it works as expected? What happens if resultsFile instance does not contain csf_coefficients array? What will be written in the TREXIO file in this case when you call write_csf_coefficient?

[q-posev]

Same comment as above about custom suffix