Backwards compatability breaks on master

[tilmantroester]

I've tried to run existing code on the current master branch and ran into a range of issues.

One repeating issue is the strict expectations that ells are integer. That is not true in practice, since any realistic measurement is going to be binned, with a window function and effective ell-label for the bin. (Formally we are working on a subset of the sphere, where the exact harmonic basis does not have integer ell).

E.g.: non-integer harmonic space scale cuts:
firecrown/likelihood/two_point.py:462: assert isinstance(self.theory.ell_or_theta_max, (int, type(None)))

This also pops up when creating the interpolation ells in LogLinearElls

Next issue is that the logic for bandpower window functions is broken again:

File [~/Codes/firecrown/firecrown/likelihood/two_point.py:464](https://vscode-remote+jupyter-002eeuler-002ehpc-002eethz-002ech.vscode-resource.vscode-cdn.net/cluster/home/tilmant/Research/LSST/baryon-challenge/notebooks/~/Codes/firecrown/firecrown/likelihood/two_point.py:464), in TwoPoint.read_harmonic_space(self, sacc_data)
    458 Cells, ells, sacc_indices, window = self.read_harmonic_spectrum_data(
    459     ells_cells_indices, sacc_data
    460 )
    461 # assert isinstance(self.theory.ell_or_theta_min, (int, type(None)))
    462 # assert isinstance(self.theory.ell_or_theta_max, (int, type(None)))
--> 464 ells, Cells, sacc_indices = gen.apply_ells_min_max(
    465     ells,
    466     Cells,
    467     sacc_indices,
    468     self.theory.ell_or_theta_min,
    469     self.theory.ell_or_theta_max,
    470 )
    471 self.theory.ells = ells
    472 if self.theory.ell_or_theta_min is not None:

File [~/Codes/firecrown/firecrown/generators/two_point.py:193](https://vscode-remote+jupyter-002eeuler-002ehpc-002eethz-002ech.vscode-resource.vscode-cdn.net/cluster/home/tilmant/Research/LSST/baryon-challenge/notebooks/~/Codes/firecrown/firecrown/generators/two_point.py:193), in apply_ells_min_max(ells, Cells, indices, ell_min, ell_max)
    191 locations = np.where(ells <= ell_max)
    192 ells = ells[locations]
--> 193 Cells = Cells[locations]
    194 if indices is not None:
    195     indices = indices[locations]

IndexError: index 20 is out of bounds for axis 0 with size 20

This can be traced back toapply_ells_min_max: it needs to use the ell-labels from the sacc data for the scale cuts, so this fails because ells here is the integer-spaced array, while Cells is the (binned) data vector

        ells, Cells, sacc_indices = gen.apply_ells_min_max(
            ells,
            Cells,
            sacc_indices,
            self.theory.ell_or_theta_min,
            self.theory.ell_or_theta_max,
        )

Next:

---------------------------------------------------------------------------
KeyError                                  Traceback (most recent call last)
File [~/Codes/firecrown/firecrown/updatable.py:192](https://vscode-remote+jupyter-002eeuler-002ehpc-002eethz-002ech.vscode-resource.vscode-cdn.net/cluster/home/tilmant/Research/LSST/baryon-challenge/notebooks/~/Codes/firecrown/firecrown/updatable.py:192), in Updatable.update(self, params)
    191 try:
--> 192     value = params.get_from_full_name(parameter.fullname)
    193 except KeyError as exc:

File [~/Codes/firecrown/firecrown/parameters.py:98](https://vscode-remote+jupyter-002eeuler-002ehpc-002eethz-002ech.vscode-resource.vscode-cdn.net/cluster/home/tilmant/Research/LSST/baryon-challenge/notebooks/~/Codes/firecrown/firecrown/parameters.py:98), in ParamsMap.get_from_full_name(self, full_name)
     96         return self[full_name_lower]
---> 98 raise KeyError(f"Key {full_name} not found.")

KeyError: 'Key Omega_c not found.'

The above exception was the direct cause of the following exception:

MissingSamplerParameterError              Traceback (most recent call last)
Cell In[9], line 33
     30 tools.reset()
     32 likelihood.update(params)
---> 33 tools.update(params)
     34 tools.prepare(ccl_cosmo)
     36 chi2 = likelihood.compute_chisq(tools)

File [~/Codes/firecrown/firecrown/updatable.py:198](https://vscode-remote+jupyter-002eeuler-002ehpc-002eethz-002ech.vscode-resource.vscode-cdn.net/cluster/home/tilmant/Research/LSST/baryon-challenge/notebooks/~/Codes/firecrown/firecrown/updatable.py:198), in Updatable.update(self, params)
    195     setattr(self, parameter.name, value)
    197 for item in self._updatables:
--> 198     item.update(params)
    200 self._update(params)
    201 # We mark self as updated only after all the internal updates have
    202 # worked.

File [~/Codes/firecrown/firecrown/updatable.py:194](https://vscode-remote+jupyter-002eeuler-002ehpc-002eethz-002ech.vscode-resource.vscode-cdn.net/cluster/home/tilmant/Research/LSST/baryon-challenge/notebooks/~/Codes/firecrown/firecrown/updatable.py:194), in Updatable.update(self, params)
    192         value = params.get_from_full_name(parameter.fullname)
    193     except KeyError as exc:
--> 194         raise MissingSamplerParameterError(parameter.fullname) from exc
    195     setattr(self, parameter.name, value)
    197 for item in self._updatables:

MissingSamplerParameterError: The parameter `Omega_c` is required to update something in this likelihood.
It should have been supplied by the sampling framework.
The object being updated was:
Omega_c

Here the issue is that ModellingTools.prepare does not take a ccl.Cosmology any more and forces the use of the CCLFactory.

After that there are issues again due to the wrong ells around apply_ells_min_max, which I couldn't hack around.

[tilmantroester]

This is the context of the baryon challenge infrastructure. E.g. https://github.com/LSSTDESC/baryon-challenge/blob/main/notebooks/run_model_predictions.ipynb