Add series date sorting to correct snap-to recency cut (#6137)

## Related PR

[6047](#6047)

## Description

The series needed to be sorted by date, so that the recency cut (five
years) could be applied when determining date of highest coverage.

## Testing

In searches such as "Demographics in Texas" or "Jobs in Texas", the
ranking and map charts should all now once again mach production in
terms of dates (see screenshots below).

## Screenshot

### Before

<img width="1401" height="714" alt="Screenshot 2026-03-26 at 8 03 59 PM"
src="https://github.com/user-attachments/assets/605297fc-d472-4be0-b6ef-7e8579df760d"
/>

### After

<img width="1387" height="712" alt="Screenshot 2026-03-26 at 8 02 38 PM"
src="https://github.com/user-attachments/assets/41988a05-b459-461a-8983-df15697b5de6"
/>

Author: nick-nlb

server/services/datacommons.py

@@ -608,6 +608,7 @@ def get_series_dates(parent_entity, child_type, variables):
       for facet_id, count in facet_counts.items():
         entity_counts.append({"count": count, "facet": facet_id})
       obs_dates.append({"date": date, "entityCount": entity_counts})
+    obs_dates.sort(key=lambda x: x['date'])
     resp_dates.append({"variable": var, "observationDates": obs_dates})
 
   return {"datesByVariable": resp_dates, "facets": all_facets}