Fornax_2019 fix #420

[adamnigel00]

Fix Fornax_2019 array time input, unit conversion, and Python 3.10+ import error

• flavor.py: add from __future__ import annotations to defer evaluation
  of FlavorScheme | None type annotations, which caused a TypeError at
  import time on Python 3.10+ due to FlavorEnumMeta.or not supporting
  NoneType (Issue 1)

• flavor_xform=AdiabaticMSW(mh=MassHierarchy.NORMAL) ->
  flavor_xform=AdiabaticMSW(MixingParameters(MassHierarchy.NORMAL)) (Issue 2) (for converting flux to events)

• ccsn_loaders.py: fix Fornax_2019._get_initial_spectra_dict to handle
  array time input correctly. Previously _get_binnedspectra was called with
  the full time array instead of a scalar, causing either a broadcast
  ValueError or silently wrong results (all time steps mapped to the same
  model index). Now finds nearest time indices upfront, uses bulk HDF5
  reads per (ebin, l, m) across unique time steps, and remaps to the full
  time array via j_inverse (Issue 3a)

• ccsn_loaders.py: fix unit conversion in _get_initial_spectra_dict —
  divide luminosity spectrum [erg/(MeVs)] by neutrino energy E to obtain
  the number spectrum [1/(MeVs)] required by flux.Container (Issue 3b)

[JostMigenda]

Type annotations

I cannot reproduce this issue on Python 3.13:

>>> from snewpy.flavor import FlavorMatrix, FlavorScheme
>>> FlavorMatrix.convert_to_flavor.__annotations__
{'flavor_out': snewpy.flavor.FlavorScheme | None, 'flavor_in': snewpy.flavor.FlavorScheme | None}
>>> type(FlavorScheme | None)
<class 'types.UnionType'>

Similarly, all unit tests (under 3.10–3.13) on the latest commit to main succeeded; the unit tests also lead to imports of snewpy.flavor. Please include reproducible instructions.

flavor_xform

When searching for flavor_xform in the diff of this PR, I get zero results. This appears to be misrepresenting the contents of this PR.

Everything else

This would take me quite some time to review thoroughly. Since the easy-to-check parts of this PR are obviously incorrect, I’m not inclined to spend more time on this and will close this as likely AI slop soon unless I hear a good alternative explanation.

[adamnigel00]

@JostMigenda Hi Jost, apologies as this PR was definitely rushed on my end, without going through proper verification so long as it worked and produced agreeable results. I will convert this to a draft as I would like to ensure this commit is reliable before any review.

With regards to flavor_xform, this was intrinsic to a script, and indeed not a part of the diff.