Integrate TROPOMI NO2 versus Pandonia

[allanjust]

In a separate TROPOMI repo @Kodiologist had previously scripted an ingestion and matchup of TROPOMI NO2 products (with irregular grids, unfortunately) and ground-based retrievals from the Pandonia network (although this may not be the only relevant ground network, it is a good starting point).

Let's integrate that ingestion here, pick some relevant predictors from the layers of the TROPOMI product (and simple derivations like focal windows) as we do for AOD, and work towards a full Earth_obs_cleaning workflow.

Ideally, we would love to improve retrievals for Tropospheric NO2. But I don't recall if we might need to work with both Tropospheric NO2 and Total Column NO2. The former is much more relevant for human health and air quality but the latter might be more directly measured by the Pandonia instruments.

Although we already cite one intercomparison, here is a NASA study in the NYC region that compares the more recent TROPOMI retrieval product with Pandora instruments (see Section 5): Judd et al AMT 2020