I'm going to wade into this to give perspective.
The algorithm that WU uses seems to be based on the temperature readings (my personal observations).
IF the temp changes rapidly, or is not updated within a few hours from the last readings then it seems that the GS is removed.
Humidity does not seem to be included in the algorithm, nor the wind readings.
Obtaining the GS does not require "5 days" of "good" readings since I and others have noted that it is achievable after as little as a few hours of re sending data after an interruption.
Even if the interruption in data is as long a few days it seems that the GS can be reinstated before 5 days after resumption.
I have noticed that the GS is removed and then the station is dropped even if the temp DROPS suddenly as in what happens when a rapidly moving front passes by.
I have brought this up with Kari and pointed out that the algorithm is such for temp increase since I suspect that it is there to weed out the PWS that are heated by sunlight.
However this should not apply to cooling. If anything this should be accounted for since it represents a more accurate portrayal of actual conditions.
The algorithm still is unchanged.
I personally think that the GS is not that meaningful. It looks nice and gives the casual observer a sense that the data observed has been sent in with some regularity (my observation again), but that is all.
The algorithm also compares your PWS readings to others around you. Woe-betide you if you are surrounded be a bunch of Netatmo PWS that are in direct sunlight, your more accurate readings are judged to be incorrect!
my $0.02 worth.