Best database to store time-series data in memory on Linux? (Embedded application)

Not sure if you have now solved this. I would suggest you read a previous answer I posted here.

One of the possible solutions from the top of my head is H2 Database Engine.

Suggest SQLite, with an in-memory option: https://www.sqlite.org/inmemorydb.html

Alternatively, can you just write to files in memory? In Linux this would be https://unix.stackexchange.com/questions/59300/how-to-place-store-a-file-in-memory-on-linux

The idea of a "database" is vague. Defined loosely as "an organized collection of data". A set of files could satisfy this. Persistence could be simply copying files to a permanent store. These files could be csv, or a serialization format like pickle https://graphite.readthedocs.io/en/latest/feeding-carbon.html#the-pickle-protocol

If you require some special access to the data, that might dictate a better fitting solution. For example, do you need to perform SQL queries?

I can load csv files into excel and plot a graph, so that seems like it would work, barring any further information.