Generalize expectation to subprobability measure

[lyonel2017]

Motivation for this change

Generalize expectation to subprobability measure

Checklist

• added corresponding entries in CHANGELOG_UNRELEASED.md

• added corresponding documentation in the headers

Reference: How to document

Merge policy

As a rule of thumb:

• PRs with several commits that make sense individually and that
  all compile are preferentially merged into master.
• PRs with disorganized commits are very likely to be squash-rebased.

Reminder to reviewers

• Read this Checklist
• Put a milestone if possible
• Check labels

[affeldt-aist]

Thanks! That looks reasonable to me. I have just added the changelog and updated the documentation.