Convenience functions for single site selection models

[nspope]

In #1082, @andrewkern suggested making the single site selection API more user-friendly. The idea is to have convenience functions that set up lists of extended events for frequently-used models like hard sweeps.

This could look something like:

contig.add_single_site(id="sweep_locus", coordinate=10000, fitness_coeff=0.1)
sweep_events = stdpopsim.ext.selective_sweep(
  single_site_id="sweep_locus",
  population_id=0, 
  start_time=1000, 
  end_time=0, 
  min_end_frequency=1.0
)

which would condition on a mutation originating in population 0 at time 1000 and reaching fixation by the end of the simulation.

Some questions are:

• What other models would be useful to support-- soft sweeps and adaptive introgression, probably? Should these be separate convenience functions, or combined into one function with more arguments?
• How much flexibility to allow in terms of allele frequency conditioning; e.g. for the hard sweep model above, should it be possible to condition on ending allele frequencies in multiple populations, rather than just the population where the mutation originated?
• To what degree should the lower-level API (lists of user-created extended events) be documented?
• Should stdpopsim.ext be removed?

[nspope]

Resolving the ext or not question should close #584. A related question is if "extended events" should be referred to as "SLiM events", as suggested in #656.