The 1872 reanalysis is too blurry here: it misses the gales the ships logged

scripts/reanalysis_fetch.py, celeste_wind.py. The 20th Century Reanalysis v3 gets the lowest Azores pressure about right (about 1001 hPa early on 25 Nov; stations 752 mm, about 1003 hPa) but its winds are weak and its spread is about +-6 m/s. At the same places and hours as 70 ship reports it gives 3.9 m/s on average against the ships' 13.9 m/s; on 1 Dec it shows 3.6 m/s where a ship logged force 12. Pressures agree within about 2 hPa on average, but the 30 Nov low is too shallow (1010-1012 vs 1005.5). Lesson: for single days at sea before about 1900, use ship logs, not the reanalysis.