Pyslim

********************

**New features**:

- Added `ts.individual_parents()`, a way to get the IDs of individual's parents
when both of them are present in the tree sequence. :user:petrelharp

- Added and documented `TSK_INDIVIDUAL_RETAINED` flag to reflect the additional
of "retained" individuals in SLiM v3.6. :user:hyanwong, :user:petrelharp

**Bugfix**:

- Modified `recaptiate` to not error with the current msprime 1.0 alpha release.

********************

********************

**Bugfix**:

- Making `.slim_generation` derive from the tree sequence's top-level metadata
had the unanticipated consequence that it could not be modified, which some
people were doing. This restores the previous behavior, but in the future,
modifying `.slim_generation` on a tree sequence will be deprecated - instead,
this should be modified in the metadata of the TableCollection.

********************

********************

**Breaking changes**:

- "First generation" individuals no longer need to be retained by SLiM to recapitate,
thanks to the "keep_input_roots" argument to simplify (new in tskit 0.3.0).
The FIRST_GEN flag and `.first_generation_individuals()` methods are now deprecated,
and if you want these to remain in the tree sequence you must explicitly Remember them.
(However, their *nodes* will remain if necessary for recapitation.)
If you wish to simplify an un-recapitated tree sequence you now can, but you must
pass `keep_input_roots=True`. This should only cause breakages if you made explicit
use of the first generation individuals, without explicitly Remembering them.

- Information about the tree sequence is now stored in *top-level metadata*,
accessible through `ts.metadata['SLiM']`. Previous interfaces remain: for instance,
`ts.slim_generation` is now redundant with `ts.metadata['SLiM']['generation']`.
This should not cause breakages, but will cause warnings where none were previously:
for instance, `pyslim.SlimTreeSequence(msprime.mutate(ts))` may throw a warning
because `msprime.mutate( )` does not preserve top-level metadata, and so SLiM-relevant
information is retrieved from provenance (as in previous file versions).

**Notable changes**:

- Switched to using tskit native encoding/decoding of metadata via schemas.
- added to conda-forge (winni2k)

**New features**:

- added `samples_only` and `population` arguments to `ts.individuals_alive_at()`
- added the `ts.slim_time()` method
- enabled dumping the reference sequence for nucleotide models

********************

********************

BUGFIX: if a tree had all first generation individuals removed
(e.g., if it had been simplified) then individuals_alive_at( ) failed.

********************

********************


This is a compatibility release, for the tskit 0.3.0 release.


**New features**:

- added has_individual_parents, a method to find individuals with all parents
are also recorded as individuals
- Provenance handling:
* added the `.slim_provenances` property to return all SLiM provenance entries
* added the `slim_provenance_version` and `parse_provenance` methods to tell if
provenance entries come from SLiM and to parse them

- documentation for recapitation with a nonuniform map by :user:TeresaPegan

**Bug fixes**:

- fixed differential time offset for tree sequences saved out in early versus late:
prior to this, mutation_at and nucleotides_at would have been sometimes wrong if the tree sequence
was saved out during late

- initialises correctly to work with tskit 0.3.0

********************

********************

**Bug fixes**:

- checks for the ability to simulate with a discrete recombination map
in the available version of msprime, and sets the default flat
recombination map in recapitate appropriately

********************