Meltpt

This release corrects a bug to allow `pyMelt` objects with multiple lithologies to be used in melt-path fitting.

No major changes. Creating neat new release to coincide with publication of M^cNab and Ball (2023, *Volcanica*, [doi: 10.30909/vol.06.01.6376](https://doi.org/10.30909/vol.06.01.6376)), with citation information, etc.

- Lots of additions/updates to the Tutorials.
- Some basic instructions for installing on Windows.
- Some changes made to allow smooth running of the code with Binder.
- Previously, correcting sample compositions for fractional crystallisation of olivine was achieved using a series of stand-alone functions and was hard-coded into the Suite class. Now, we have collected these functions into a single class, "BacktrackOlivineFractionation", which is instantiated then passed to Suite. This change will facilitate future incorporation of alternative backtracking schemes.
- Some minor changes to the way bounding potential temperatures for a suite of pressure/temperature data are found.

* The melting parameterisations used in **meltPT**, via **pyMelt**, use a
quadratic function for the solidus, so that for large potential temperatures
no melting path can be calculated. To avoid optimisation algorithms
attempting to test such high potential temperatures, we included an
arbritrary upper bound on potential temperature. In this new version, rather
than being set abritrarily, the upper bound is chosen based on the properties
of the mantle object passed to the optimisation routine.

* A bug has been fixed which caused sample-composition backtracking with a
constant value for the partition coefficient to fail in verbose mode.

* A discrepency between the script Examples/Tutorials/Tutorial_1.py and the
version rendered in the documentation has been fixed.

Releasing version 1 of meltPT.

Final preliminary test release?