Long delays until position is acquired

I noticed that UnifiedNLP has really long delays until a changed location is being reported. E.g. a weather widget might stay for several hours (some times more than 12h) on the same location until it notices a change in location. Especially annoying is that Tasker (the automation app) profiles based on network location (without GPS) are extremely unreliable: Tasker cell contexts, i.e. profiles based on the ID of the current cell tower, work perfectly, the same profile with UnifiedNLP and location context has a delay of 15+ minutes to hours. And that even if the phone is definitely logged into cell towers which should trigger the context.

I also can have Mozilla stumbler running in parallel. As this programm shows the current location as computed by Mozilla NLS, I see that the correct position is available, but UnifiedNLP does not report it until much later.

I use a Samsung S7 with different custom MM-ROMs, and MicroG-UnifiedNLP with different backends (Mozilla and OpenCellID). It somehow seems as if UnifiedNLP is kind of dozing and not updating its position info. Maybe some power saving mechanism of Android MM is interfering with the continuous work of UnifiedNLP, but I have not yet found a solution.

No source files or tests are named. Start by reproducing the delay on a Samsung S7 running Android MM with Mozilla or OpenCellID, comparing the position shown by Mozilla Stumbler with the updates reported by UnifiedNLP. Done means location changes are reported promptly enough for weather widgets and Tasker network-location contexts.

Written by the indexing model from the issue text.