Psifr

This version adds support for block lag CRP analysis, which measures lags between blocks of serial positions. It also improves the flexibility and consistency of some utilities.

* Support and unit tests for block lag CRP analysis.
* Support for changing the lag field in lag-CRP analysis and determining whether counted possible transitions are transition-unique or not.
* Change to list splitting for better consistency. Now item queries are applied to both study and recall events instead of just study events, except in the context of a transition analysis.
* Added documentation on item and transition filtering.

This version fixes a problem with the `count_unique` option in the distance CRP analysis. It also adds support for counting non-unique lags, which will be necessary for implementing a block CRP analysis.

* Using the `count_unique` option for `psifr.fr.distance_crp` will now only increment each distance bin as possible once per transition. Previously each unique possible distance value was recorded for each transition, rather than each unique possible distance bin.
* A `count_unique` option has been added to `psifr.transitions.count_lags` to handle cases where multiple possible transitions may have the same lag.
* A configuration file for Zenodo has been added to set project information automatically.

This version adds more documentation and does work behind the scenes to better organize the code and make writing extensions simpler.

* Refactor all unit tests to use pytest instead of unittest.
* Reorganize unit tests to be more consistent and to remove redundant fixtures.
* Modernize the build system using updated standards.
* Reorganize the core analyses to separate measures, transitions functions, and outputs functions. The high-level `psifr.fr` API remains the same, so this change only affects development of new analyses.
* Make code style consistent by reformatting using Black.

This release is primarily to trigger archiving on Zenodo. It will also be the tagged version for the JOSS paper.

This update adds:
* A quickstart section to the README illustrating how to create and plot a serial position curve.

This version adds more documentation, including additional explanation of how to interpret analysis results. A paper under review at JOSS has also been added, which gives background on the analyses implemented in the package and differences from other similar packages.

Other changes:
* Improved figure legends on raster plots.
* Fixed data filtering when using scalar float values.
* Improved automated build settings.
* Improved test coverage of data handling and plotting.

This update adds:

* A utility for loading sample data.
* An option to set plot marker size.
* A fix to a problem when using a transitions test that only references the previous item.
* Improvements to the documentation.