Heat spells

[aulemahal]

Pull Request Checklist:

• This PR addresses an already opened issue (for bug fixes / features)
  • This PR fixes #xyz
• Tests for the changes have been added (for bug fixes / features)
  • (If applicable) Documentation has been added / updated (for bug fixes / features)
• CHANGELOG.rst has been updated (with summary of main changes)
  • Link to issue (:issue:number) and pull request (:pull:number) has been added

What kind of change does this PR introduce?

• 3 new indicators heat_spell_frequency, heat_spell_max_length and heat_spell_total_length. The "heat_wave" indicators test that a bivariate condition on each day and then find consecutive runs of a minimum length. The new functions adhere to xclim's concept of "spell" : the condition is tested against a rolling N-day window and if it is fulfilled all days of the window are considered part of the spell.
• New generic function spell_mask that encapsulate the spell notion. It returns a boolean mask of the days part of spells. The code has been added two features:
  • Multivariate : input data can be a list (and so must be the thresholds). The way we check for the conditions across variables is decided by new argument var_reducer ('all' or 'any').
  • Weights : if the spell stat is a mean, weights can be given (used by the INSPQ, but not for the actual PC indicators).
• Indicator parameters can now be assigned new names. See examples in how the heat spell indicators are created. We could already do something similar with data variables. The internals of Indicator have changed accordingly (sorry).

Does this PR introduce a breaking change?

No.

Other information:

I kinda decided that "heat" meant "tasmax and tasmin".

[coxipi]

approved

[coveralls]

coverage: 89.114% (+0.01%) from 89.1%
when pulling 3d276a0 on heat_spell
into 7acc410 on main.