Faps

Added function to return human readable results about the top candidate fathers
for each individual offspring in the data set. This essentially a summary
function for sibshipCluster.posterior_paternity_matrix().

- Added method `paternity` to sibshipCluster objects to return a table for that
halfsib array only giving the IDs of the mother, offspring, and the N candidate
fathers with the highest posterior paternity, along with those probabilities.
- To do this I needed to update sibshipCluster to inherit offspring names from
paternityArray objects.
- Renamed sibshipCluster.prob_paternity to the more intuitive
`posterior_paternity_matrix`

Added method `.posterior_mating` to make posterior draws of offspring from a sibshipCluster object to return a data frame of plausible fathers, along with the proprtion of pollen coming from each father. This can be run over multiple sibshipCluster objects with the general funciton `posterior_mating`. These functions replace `mating_events`, which had a very obscure output. Fixed a bug in the tests for these functions that was incorrectly splitting up the mothers and giving weird results.

Major update to how FAPS calculates paternityArrays to handle 3-d arrays of genotype probabilities so that I can properly integrate out uncertainty in genotype (for biallelic SNPs). `genotypeArray` objects now take the argument `geno_probs` as their second argument, giving probabilties that each locus is 0, 1 or 2. The previous implementation can still be called by passing `integration = 'partial` to `paternity_array(`.

Updated simulation functions:

1. Corrected bug in make_generation that called range, because this has changed in python 3.
2. Removed ipywidget progress bar from make_power and replaced with tqdm, which is much lighter.
3. Removed the package dependency on ipywidgets.

Changed `genotypeArray` to use only `np.nan` for missing data, replacing -9. Also replaced all references to the use of -9 in other functions, except for `export_to_colony(`, which converts np.nan to -9, then adds 9 (zero is Colony's symbol for missing data).

Major update to ensure compatibility with python 3. In addition, version 2 deals with multiple genotype and paternity arrays as dictionaries indexed by maternal family (replacing lists in version 1).

Various small changes since publication.

Version used to create simulations and analyses in associated with the FAPS paper "Efficient inference of paternity and sibship inference given known maternity via hierarchical clustering" in Molecular Ecology Resources (https://doi.org/10.1111/1755-0998.12782). Support and documentation for:
- handling genotype data
- creating or importing paternity arrays (G matrix from the paper)
- clustering into sibships, and investigating sibship structure
- drawing posterior samples of mating events.