Kasper Peeters

At the moment evaluate tries to write out all indices. Provide option to restrict to a single index set.

E.g. `examples/ref_programming.cnb` or in fact any notebook with error or verbatim cells. They do get typeset by TeXEngine, but somehow do not appear on the screen.

Mostly handled in `is_scalar_function`, but we should do it the other way around: only wrap in `components` if necessary, then we also do not need to unwrap afterwards.

See Q&A https://cadabra.science/qa/1155/eliminate_vielbein-and-the-kroneckerdelta for details.

See https://cadabra.science/qa/1284/extending-zoom-function ; we need a way to specify that the pattern `a Q??` should match even when `Q??` is empty, or there should be a way to specify multiple...

The `unwrap` algorithm only handles simple dependencies, not dependencies on e.g. `\dot{a}` or other composite objects.

The online reference manual still misses sections from the 1.x reference manual.

Cadabra's notebook interface can already be used to work with Sympy, but it still requires installing the entire Cadabra software. It may be useful to separate the interface from the...

With derivatives in anti-commuting directions, a partial derivative with multiple indices should be anti-symmetric, not symmetric.

The `can_swap` functions silently ignore that differential forms may be matrix-valued and hence not pick up the 'standard' sign; see https://cadabra.science/qa/1276/manipulating-matrix-valued-differential-forms for an example. Need at least `SelfAntiCommuting` and `SelfNonCommuting`...