Radioactivedecay

- Latest SymPy release (v1.9) changed internals of Rational / Matrix objects. This breaks loading
of SymPy <=1.8 pickle objects when using SymPy v1.9. `radioactivedecay` now packages SymPy pickle
files appropriate for SymPy <=1.8 and for SymPy >=1.9. It checks which SymPy version is in the
local environment before choosing the correct files to load (fixes issue 50 and failure of
`radioactivedecay` v0.4.4 to build on the conda-forge feedstock).
- Added support for Python v3.10 (PyPI only, not yet available via conda-forge).
- Move responsibility for file I/O from `DecayData` class initiator into separate function called
`load_dataset()`.

- Fix docstring typo, Inventory `plot()` method (48).

- Refactored DecayData class to store progeny, branching fractions and decay modes data in
separate numpy arrays.
- Use np.array_equal() for checking equality of numpy arrays.
- Fix bug which caused scipy C and C^-1 matrices to be unnecessarily large.
- Added tests to check shape of scipy C and C^-1 matrices from icrp107_ame2020_nubase2020 dataset.
- Use idx as variable name for index of for loops.

- Fixed bug in `InventoryHP` meaning `decay()` gave some incorrect results.
- Hard-coded Avogadro's constant to avoid tests failing with older versions of SciPy.
- Updated notebooks to be consistent with v0.4.0+.

- Added ICRP-07 and AMDC license files into MANIFEST.in. Includes these files in PyPI package.

masses, moles and numbers of atoms when creating inventories, and also methods so inventories can
report their contents in terms of these quantities, as well as mass or atom fractions (35). Mass
conversions use atomic mass data from the Atomic Mass Data Center (AMDC) by default.
- To enable SymPy high precision calculations throughout unit and quantity conversions, there is a
new `InventoryHP` class (high-precision inventory class). This behaves the same as the normal
precision `Inventory` class. The old `Inventory.decay_high_precision()` method is now deprecated -
use `InventoryHP.decay()` instead for high-precision decay calculations.
The number of significant figures for high precision decay calculations is now specified via the
`InventoryHP.sig_fig` attribute (default is 320) rather than as a parameter to the `InventoryHP`
`decay()` and `plot()` methods.
- Added new `cumulative_decays()` method to the inventory classes. This calculates the total number
of atoms of each radionuclide that decay over the decay time period.
- Added support to specify nuclides using canonical ids (36).
- Documentation updates for all new and modified functionality.
- LICENSE file split into separate files for `radioactivedecay`, ICRP-107 decay data, and AMDC
atomic mass data (38).
- Added parsing of AME2020 and NuBase2020 atomic mass and isomer excitation energy data to ICRP-107
dataset Jupyter notebook. Added stable nuclides to decaydata.npz file, C and C_inv matrix SciPy
and SymPy files in ICRP-107 (33 & 34).