replace ARM GHI with Best Estimate GHI

[wholmgren]

• [ ] Closes #xxxx .
• [x] I am familiar with the contributing guidelines.
• [x] Tests added.
• [ ] Updates entries to docs/source/api.rst for API changes.
• [x] Adds descriptions to appropriate "what's new" file in docs/source/whatsnew for all changes. Includes link to the GitHub Issue with :issue:`num` or this Pull Request with :pull:`num`. Includes contributor name and/or GitHub username (link with :ghuser:`user`).
• [x] New code is fully documented. Includes numpydoc compliant docstrings, examples, and comments where necessary.
• [x] Maintainer: Appropriate GitHub Labels and Milestone are assigned to the Pull Request and linked Issue.

[wholmgren]

Warning added in 32ceb99 fired as expected:

2021-09-07 16:38:44,046 WARNING Duplicate index values in sgpmetE9.b1. Keeping last.

The deduplicator is needed at this level of the code because reference_observations/arm.py concatenates the irradiance and the meteorological data along the columns, and this only works if there are no duplicates in either dataframe. Debatable whether the deduplicator should be here or in reference_observations/arm.py. Going with the most straightforward approach for now.

The first deduplicator added in a268a94 did solve the problem for ARM sites that have duplicates in both the irradiance and the met data. Sentry logs indicate that this problem has also occurred with at least a MIDC site, so let's keep these lines too.