Clean up size_function semantics.

[grahamgower]

Closes #262.

[codecov]

Codecov Report

Merging #278 (433f9e2) into main (69d4f0d) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main     #278   +/-   ##
=======================================
  Coverage   97.51%   97.51%           
=======================================
  Files           8        8           
  Lines        1410     1410           
=======================================
  Hits         1375     1375           
  Misses         35       35           

Impacted Files	Coverage Δ
demes/demes.py	100.00% <ø> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 69d4f0d...433f9e2. Read the comment docs.

[apragsdale]

This is looking good - I like the documentation you've added. Just one question about always defaulting to "exponential" here, but maybe I'm missing something..

[jeromekelleher]

This looks good. Reading through the thread, here's my suggestion:

1. If start_size == end_size and no size function is specified, default to "constant"
2. If start_size != end_size and size_function == "constant" raise an error

That is, hard equality rather than math.isclose. It's hard to imagine cases where the values would differ if the size function really was constant and it's not an error.