Cwinpy

Major changes for this release:

- Allow the use of a "transient-continuous" signal model (!108)
- Fix how the minimum Bayesian Block chunk length is set in `HeterodynedData` (!114)
- Use SciPy filter functions for performing filtering during the heterodyne stage (!112)
- Move `likelihood.py` into the `pe` submodule (!116)
- Allow `pulsarfiles` configuration file option for `cwinpy_pe_pipeline` that is consistent with the `knope` and `heterodyne` pipeline (!117)
- Fix bug that now mean that PE will work for frequency/frequency derivative parameters (!120)
- Add the ability to use the Einstein Telescope (ET) detector ASD for simulated noise (!121)
- Exclude the log likelihood and log prior from the `Plot` posterior plots by default (!126)

The release has major changes, including some that are backwards incompatible. The major changes are:

- Allow pipelines to be given ISO format start and end dates as well as GPS times (!106)
- Greatly speed-up the running median calculation for `HeterodynedData` (!93)
- Greatly speed-up the Bayesian Blocks calculation for `HeterodynedData` (!94)
- Fix bugs that now allow non-GR parameters to be estimated (!99, !100)
- Correctly deal with input frame data that is 32-bit floats when heterodyning (!98)
- Add pipeline to perform sky-shifting analysis (!72)
- Set the default dynesty sampler method to be `rslice` (!91)
- Fixes to allow running pipelines on the OSG (!70, !101)
- Switch from using pycondor to HTCondor Python package (!89)
- Remove dependencies on bilby_pipe package (!90)
- (**Backwards incompatible**) Change `_dag` suffix on pipeline scripts to `_pipeline` (!71)

Changes for this release:

- Fix bug when combining multiple heterodyned time series during heterodyne pipeline (!69)

Changes for this release:

- Fix bug in resample rate usage when resuming a heterodyne analysis (!68)

Changes for this release:

- Add the `cwinpy_knope` and `cwinpy_knope_dag` scripts for running the full heterodyne and PE pipeline (and generate HTCondor DAGs) (!60)
- Move from using the `PulsarParametersPy` class from LALPulsar to a version within CWInPy itself (!66)
- Allow use of TEMPO2 (via libstempo) for calculating the phase (!63)
- Add `Plot` class to allow plotting of various posteriors (!62)
- Allow parameter estimation to be performed both coherently for multiple detectors and for the individual detectors when submitting a PE DAG job (!61)
- Use GWOSC data find server for finding local paths of CVMFS data rather than hardcoding these (!65)

Note: several of of these changes is are major updates and may break some backwards compatibility.
There may still be some minor bugs in the implementation and there are more comparison tests to
perform, but the avoid further unwieldy MRs these have been put into a release. This release also
requires a development version of LALSuite, so a Conda installable release of this version will not
be available.

Changes for this release:

- Add command line interface for data heterodyning, including generating HTCondor DAGs (!28)
- Add the ability to merge `HeterodynedData` HDF5 files (!51)
- Allow heterodynes to use pulsar ephemerides from the ATNF pulsar catalogue (!53)
- Switch the default save format of parameter estimation outputs to be HDF5 files rather than JSON (!54)
- Add heterodyned signal simulation class into CWInPy rather than using LALSuite version (!46, !55)
- Add a histogram distribution to the hierarchical analysis (!48)
- Fix how the KDE bandwidth is calculated in the hierarchical analysis (!49)
- Change the way packaging/versioning of the code is done (!52)

**Important note**: HDF5 files created from `HeterodynedData` objects using earlier versions of
CWInPy will no longer be compatible with v0.6.0 and will fail to be read.