QC of SPN1

Besides the negative values reported in #57, then the SPN1 data seems to be in good condition. The plot below (GHI>50) is without applying any QC checks or manually flagging besides
> "2019-05-27 09","2025-03-31","sun_total;sun_diffuse","remove","SPN1 was taken down"
> "2017-12-21 00","2018-01-11 09","sun_total;sun_diffuse","remove","Stuck at high value"

- [ ] Apply BSRN limit checks


```python
dfp = df_qc[df_qc['sun_total'] > 50]

fig, ax = plt.subplots()
plt.scatter(
x=dfp['sun_total'] / dfp['ghi_extra'],
y=dfp['sun_diffuse'] / dfp['sun_total'],
c=solpos.loc[dfp.index, 'apparent_elevation'],
s=0.5, alpha=0.2)

ax.set_xlabel('sun_total / ghi_extra')
ax.set_ylabel('sun_diffuse / sun_total')
```

Here is a histogram of the SPN1 irradiances:
```python
df_qc[['sun_total','sun_diffuse']].plot.hist(log='y', bins=100, alpha=0.5)
```


Start by locating the SPN1 data-processing or QC entry point that produces df_qc, then review how the existing manual removals for the two listed periods are applied. Implement the BSRN limit checks for sun_total and sun_diffuse, verify the resulting data and plots, and mark the checklist complete when the checks are applied.

Written by the indexing model from the issue text.