GitLab

Want to allow for the use of arbitrary node beam parameter properties as target
variables in minimisation function routines. This requires a change to how tracing
results are initialised and stored as the beam parameter associated with the target
variable must remain as a valid reference during the beam tracing - i.e. no overwriting
of BeamParam instances in Model.__last_trace.data, only resetting of the q property.

as a single file

This file installs all of the dependancies
including SUITESPARSE.

You can install it into a new enviroment using
`conda env create -n finesse3 -f enviroment.yml`

OR you can update your exisiting enviroment to
include the finesse dependancies using
`conda env update -f enviroment.yml`

Then you need to run
`python setup.py develop`
to install finesse.

adding animated example of beam tracing execution to documentation and improving ABCD matrix manual section images

We previously had some documentation files which displayed all the attributes and methods for
several classes on the same page - which looked ugly and wasn't very well organised. So I've
altered this such that each class has its own page in the documentation for better readability
and ease of use - these classes are then referenced in autosummary toctrees on the pages they
used to be displayed on.

As part of this documentation update, I've switched over from using the autogen.sh script to
a new autogen.py file which is then executed when running the documentation - as recursively
calling sphinx-autogen through the required directories is easier with a Python script. Hopefully
this switch-over won't be an issue when the documentation is built for official use in the future.

Note that the check_missing.py script needs to be updated for compatibility with this new API
documentation structure.

changing __str__ overloads of TracerOut and OpticalPath classes to return nicely formatted tables of data

Integration tests directory now contains some simple tests for beam tracing (laser -> mirror and
Fabry-Perot). Added more mirror and beam splitter ABCD unit tests.

Added data_qx and data_qy properties to TracerOut class for obtaining a dictionaory of just the tangential
and sagittal parameters from a beam trace run for each node, respectively.

Simulations are now almost up and running for more advanced setups like
the aLIGO `design.kat` file, though it currently produces a singular
matrix. If we ignore that, however, the simulation runs and some
(extremely vaguely) sensible outputs are generated.

Currently puts don't handle variables of any sort, only numeric
constants, but xaxis can already be used for basic stuff.

updating check_missing script to show all functions and classes missing docstrings / not in doc files

Building with Conda on windows is likely impossible, as visual studio
complains about different compiler versions used to build libklu.lib and
the main CMatrix.lib - I think pre-built binaries will be the easiest
option here.

Removed Model.full_path so that Model now has just one method for finding
the path between two optical nodes - Model.path. This now returns an object
of type OpticalPath which acts as a wrapper around the underyling (node,
component) data.

This change allows for much nicer formatting of a path when printing an
OpticalPath instance.