is it normal?

Yes, for CEP2 and -all- VINA based sciences such as FAHV, SN2S, DSFL you will see 2 or more processes listed. Only 1 [the worker] for each task uses CPU time continuously, when BOINC is allowed to run.The checkpointing you can control [for all but CEP2] by increasing the "Write to Disk at most" setting, which is default set to 60 seconds. 5 minutes (300 seconds), is a good number to set to. For devices that hardly ever get booted you can enter as high a preference as you like [think newer clients have no limit]. Remember though that if you boot / restart a client all progress is lost since last checkpoint write, which you see in the event log, if you've specified in the cc_config.xml to record the checkpoint debug output. Else look in task properties to see when last one was written. Alternately the Task Manager process view, all users, shows how long the current [forked] job segment has been running. This is the same as telling how long ago the last save took place.

edit: Add to last line

Given this is Android forum, on checkpointing and write to disk control, there is currently actually no control to increase the interval. This I've proposed to the developers in this testing forum thread: https://groups.google.com/forum/#!forum/boinc-android-testing

Why the task reverts I do not know, but if there is BOINC for Android option to "leave application in memory when suspended", you need to set that. Whilst, the frequent checkpointing prevents major progress loss, maybe the reason why it is done that way. No idea though why things for Android were done the way they were done [small internal RAM consideration, absent ability to use SSD storage?], also because users have a concern over the frequent writing to flash memory.

@ the techs: I'm not very impressed with that place. Posted on the 19th, no response, not even an acknowledgement the issue is being looked at. Hot potato?

You have no errors, all valid, we explained the forking of the process and why it sometimes could go back in progress [task interrupted in-between checkpoints]. Think we answered all questions. If there is another issue then start a new topic.