2.3.2

Various improvements and fixes, and a new default Jupyter kernel:

• Fixed bug in pattern matching logic used in integrate_by_parts; see this Q&A post.
• Fixed bug which would prevent assignment to node multiplier from Python.
• Tab-completion on command line and in the notebook.
• Install the cadabra2 python module in a standard location (Atri). You may still need to set PYTHONPATH if you are on Debian/Ubuntu.
• Fixes for the Mathematica scalar backend.
• Make expand_delta much faster when the Kronecker delta is contracted with objects which have anti-symmetry.
• Make meld work correctly with non/anti-commuting objects (Dominic).
• Make numbered indices like a1 display with a subscript automatically a<sub>1</sub>.
• Provide pickle functionality for expressions (Dominic).
• Fixes for compatibility with newer SymPy versions.
• Added a 'find' function in the notebook, to search input cells.
• New Jupyter kernel (enabled by default) which does not require Xeus; you can now get access to Cadabra on Jupyter on all supported systems, not just Conda (Fergus).
• Various improvements and fixed for handling and display of Young tableaux.
• Added ExNode::ex() to obtain an Ex object from an ExNode iterator.