More descriptive error handling for API calls

There should be better error responses when using cudaq_solvers.create_molecule and any other code that uses API calls to execute code. When executing the following, the user just gets a blanket error that isn't specific about what went wrong on the pyscf call

from cudaq_solvers import create_molecule

xyz = [('H', (0., 0., 0.)), ('H', (1.0, 0., 0.))]
molecule = create_molecule(xyz, 'sto-3sg', 0, 0)

---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
Cell In[34], [line 4](vscode-notebook-cell:?execution_count=34&line=4)
      [1](vscode-notebook-cell:?execution_count=34&line=1) from cudaq_solvers import create_molecule
      [3](vscode-notebook-cell:?execution_count=34&line=3) xyz = [('H', (0., 0., 0.)), ('H', (1.0, 0., 0.))]
----> [4](vscode-notebook-cell:?execution_count=34&line=4) molecule = create_molecule(xyz, 'sto-3sg', 0, 0)

RuntimeError: HTTP POST Error - status code 500: : Internal Server Error

when the problem is just that an invalid basis set ('sto-3sg' rather than 'sto-3g') was passed. Similarly, if the xyz is changed such that it is syntactically correct but not what pyscf expects then the error will be the same blanket internal server error.

Start at the cudaq_solvers.create_molecule entry point and trace the API calls used to execute code. Reproduce the invalid basis-set and malformed-xyz examples, then make the resulting errors identify the underlying PySCF problem rather than only reporting HTTP 500; the issue names no files or tests.