persistence interval label issue

[wholmgren]

It appears that the reference probabilistic persistence time of day forecasts are not accounting for the desired interval label = ending.

I'm not 100% sure but I think the issue may be that the following lines need label=closed in addition to closed=closed:

solarforecastarbiter-core/solarforecastarbiter/reference_forecasts/persistence.py

Lines 173 to 175 in fa8fb99

	# average data within bins of length interval_length
	persistence_quantity = obs.resample(interval_length,
	closed=closed_obs).mean()

solarforecastarbiter-core/solarforecastarbiter/reference_forecasts/persistence.py

Line 479 in fa8fb99

obs = obs.resample(interval_length, closed=closed).mean()

solarforecastarbiter-core/solarforecastarbiter/reference_forecasts/persistence.py

Line 590 in fa8fb99

obs = obs.resample(interval_length, closed=closed).mean()

So that suggests the problem is not just in the prob persistence time of day.

Note the presence of label=closed in the index function:

solarforecastarbiter-core/solarforecastarbiter/reference_forecasts/persistence.py

Lines 310 to 312 in fa8fb99

	# resample sub-interval reference clear sky to observation intervals
	clear_ref_resampled = clear_ref.resample(
	observation.interval_length, closed=closed, label=closed).mean()

[wholmgren]

In line 20 of workshop/reference_forecasts.ipynb we use label='ending', so similar but different problem. Subtle but real changes:

[wholmgren]

As documented in new comments in #702, only persistence_probabilistic_timeofday is failing to account for the interval label - the functions in persistence.py are actually fine.