Explore the sites with unusually high or low temps

Holly was using our GLM output data and seeing some cases where the temperatures do not seem reasonable - 8 deg C in Dec, etc. She provided some examples of these weird data via Slack message in general on 3/3/2022. She also shared some figures in our 3/3/2022 meeting docs documenting some of these outliers. We should look into this more but some initial potential ideas for the cause:

- Using older GLM version (see idea about updating version in shifter image in #5)
- N/A cells (~100 lakes) (we have a fix but haven't re-run the GLM models, see #32)
- Previously noticed that layers in the model flattened into a single layer. Simulation ran but produced bad results so wasn't flagged.

One thing we wanted to do was explore if any of the sites Holly shared are similar to the ones we are seeing produce errors (see #17). Here are a few of the sites/years she shared via slack on 3/3/2022:

```
site_id,year,mean_surf_dec
nhdhr_0cb9b9cf-3f4b-4061-9608-9f965f202b85,2057,9.336258292927424
nhdhr_117649747,2057,8.003453159776221
nhdhr_118603427,2057,8.570490833097425
nhdhr_118874139,2057,8.906114791135298
nhdhr_118874243,2057,8.067051710676578
nhdhr_118874447,2057,8.192774192523245
```

```
site_id,year,mean_surf_jun
nhdhr_114541955,2000,-1.436453848511092
nhdhr_120018433,2044,5.822018574085484
nhdhr_120020387,2098,8.176853422082687
nhdhr_120020388,2098,2.575703962717457
nhdhr_120020568,2088,-0.8596796607315065
nhdhr_121544759,2084,9.979001749159783
```

Start with the listed site/year examples and compare them with the error-producing sites referenced in #17. Review the hypotheses about the GLM version, N/A cells, and flattened layers, including related issues #5 and #32. Done means identifying whether the examples share a cause and documenting the affected sites or years and the likely source of the anomalous temperatures.

Written by the indexing model from the issue text.