Major feature release

Major feature release adding documented, stable support for simulating selection (via SLiM),
as well as new species and models.

Bug fixes:

• Incorrect scaling of the DroMel/LognormalPlusPositive_R16 DFE led to negative selection
  coefficients that were too large (many on the order of -1e3); scaling was fixed for the
  log scale (@clararehmann, #1699)

• Factor of two error fixed in HomSap/Gamma_K17 DFE (@RyanGutenkunst, #1478)

• Discretization for exponential growth using the SLiM engine led to the final size
  in growing/declining populations not matching (by a small amount) what is listed
  in the model; improved discretization scheme (@petrelharp, #1622)

• Updates to SLiM support: updated the active flag in the SLiM code to be integer.`

Breaking changes:

• The 2018 Browning et al HomSap demographic model previously named
  "AmericanAdmixture_4B11" is now named "AmericanAdmixture_4B18"; the old name
  works but throws a deprecation warning. (@petrelharp, #1603)

• The time_units attribute of tree sequence metadata is now set to "generations"
  for SLiM output, avoiding the "time units mismatch warning".
  (@nspope, #1567)

• Previously, Contigs specified with left and right arguments produced
  simulations having coordinates shifted so they were relative to left. This
  is no longer the case: coordinates in resulting simulations retain the
  original chromosome's coordinates. As a result, regions outside [left, right) will now have missing data. To produce simulation output with the
  previous behavior, use the .trim() method of the resulting TreeSequence.
  So, the Contig's .original_coordinates attribute is now called
  .coordinates, and several methods (e.g., Contig.basic_contig()) now take
  left and right arguments as well as length. Finally, arguments such as
  relative_coordinates=True/False to Contig.dfe_breakpoints() are no longer
  necessary and are deprecated. (@petrelharp, #1570)

• To add the possibility for a relationship between dominance and selection
  coefficient, now each stdpopsim MutationType might have more than one
  underlying SLiM mutation type; so, where this is recorded in top-level
  metadata (under ts.metadata["stdpopsim"]["DFEs"]) is now a list
  instead of a single value. This will not affect anyone who is not
  parsing the metadata related to DFEs.

• SLiM extended events and selective sweep infrastructure have been
  moved from the stdpopsim.ext namespace into stdpopsim proper.

• The length_multiplier option to Species.get_contig is deprecated and
  prints a warning. The options left and right should be used to truncate a
  contig, instead. (@nspope, #1605)

• Added assembly_source and assembly_build_version properties to
  Genome objects, so that different genomes could be based on
  different sources (previously, all were with reference to the Ensembl
  release 103). (@andrewkern, #646

New features:

• Relationship between dominance and selection coefficient:
  Added the dominance_coeff_list argument to MutationType, allowing
  for DFEs with a discretized relationship between h and s.
  (@petrelharp, #1498)

• Model specific recombination rates:
  Added the recombination_rate attribute to demographic model,
  allowing for model specific recombination rates in addition to
  the species default rates.
  (@gregorgorjanc, #1591)

• Added support for the SLiM engine on Windows. (@petrelharp, #1571)

New species:

• Mus musculus (@peterdfields, #1437).
  QC'd by @igronau, #1454.

• Orzya sativa (@ornobalam, #1453).
  QC'd by @minesrebollo, #1461

• Phocoena sinus (@igronau, #1514).
  QC'd by @ckyriazis, #1538

• Gorila gorila (@ChristianHuber, #1517)
  QC'd by @andrewkern, #1701

• Rattus norvegicus (@kevinkorfman, #1700)
  QC'd by @andrewkern, #1706

• Sus scrofa (@AprilYUZhang, #1672)
  QC'd by @gregorgorjanc, #1689

New DFEs:

• Generic "uniform" DFE (@petrelharp, #1492)

• HomSap/LogNormal_H17 (@RyanGutenkunst, #1480)
  QC'd by @clararehmann, #1698

• HomSap/Mixed_K23 (@chriscrsmith, #1505)
  QC'd by @clararehmann, #1696

• PhoSin/Gamma_R22 (@igronau, #1547)
  QC'd by @ckyriazis, #1560

• AraTha/Gamma_H18 (@chriscrsmith, #1324)
  QC'd by @clararehmann, #1647

• HomSap/GammaPos_H17 and DroMel/GammaPos_Z21 (@petrelharp, #1656)
  QC'd by @clararehmann, #1694 & #1695

New demographic models:

• MusMus/DomesticusEurope_1F22 (@peterdfields, #1485)
  QC'd by @igronau, #1531

• MusMus/MusculusKorea_1F22 (@peterdfields, #1485)
  QC'd by @igronau, #1531

• MusMus/CastaneusIndia_1F22 (@peterdfields, #1485)
  QC'd by @igronau, #1531

• OrySat/BottleneckMigration_3C07 (@ornobalam, #1453)
  QC'd by @minesrebollo, #1466 and @petrelharp, #1524

• PhoSin/Vaquita2Epoch_1R22 (@igronau, #1526)
  QC'd by @ckyriazis, #1538

• CanFam/EarlyWolfAdmixture_6F14 (@agladstein, #1632)
  QC'd by @gregorgorjanc, #1644

• BosTau/HolsteinFriesian_1B16, BosTau/Angus_1B16, BosTau/Fleckvieh_1B16, and
  BosTau/Jersey_1B16 (@gregorgorjanc, #1593)

New annotations:

• PhoSin exons and CDS (@chriscrsmith, #1520)

New QCs:

• DroMel/Gamma_H17 (@clararehmann, #1668)

• DroMel/LognormalPlusPositive_R16 (@clararehmann, #1699)

• HomSap/Gamma_K17 (@clararehmann, #1687).

• MusMus/Gamma_B21 (@clararehmann, #1669)