For each iterator the output will now also output the time took for the
iteration and the aggregated time from the start of the solver. Also
some information, which doesn't change are written once and not again.

Using this CartesianIndices, you can create a n-dimensional recetengular
shape and iterate over the region. This should be quite a powerful but
easy to use abstraction over the volume.

The provided free functions, can be used to visit neighbouring voxels
from a given position.

This is mostly as I've encountered the bug described
https://github.com/doctest/doctest/issues/473 this bump fixes that for
me and hopefully everyone using a new glibc version.

See changelog for version 2.4.7
https://github.com/doctest/doctest/blob/master/CHANGELOG.md#247-2021-12-10

The read and write free functions decide based on the extension given,
what format to read and write, making it a little more convenient to
quickly switch output and input formats. So it's only necessary to
change the string, rather than the string and the namespace prefix.

Add a constructor taking a format string and arguments, which are passed
to fmt::format. This makes it quite convenient to pass arguments to
exceptions, i.e. `throw Error("value should be 0, but is {}", value);`.

- use APSInt::toString(llvm::SmallString) everywhere
- create() instead of construct options parser for LLVM>=13

otherwise we will generate a module config with name `DEPENDENCIES`.

David Frank authored Dec 27, 2021 and Tobias Lasser committed Jan 07, 2022

Instead of first creating a WLSProblem and passing that to the
constructor, you can now pass the projector and the sinogram to the
problem. The constructor initializes a very basic form of tikhonov
regularization, for other forms, the other constructors should be used.

David Frank authored Dec 27, 2021 and Tobias Lasser committed Jan 07, 2022

Instead of constructiong `L2NormPow2(LinearResidual(A, b))`, you can
now simply write `L2NormPow2(A, b)`.

This commits adds some default hooks provided by pre-commit, such as
hooks to check for too large files, or to prevent committing to master,
or files with unresolved merge conflicts.

Check the contribution guide on how to use the hooks

Some styling choises of our in-code documentation just doesn't looks
nice in the output of the docs. Here, I've mostly fixed the '@author'
tag. Instead of having each author in their own tag, just have the tag
and then a list starting in the next line.