Sybenzvi/csv to sql

[sybenzvi]

This PR addresses issue #98 by doing the following:

• Removes the 24-hr and 7-day CSV alert caches and logs.
• Implements connection to an SQLite disk file using sqlalchemy
• Automatically removes alerts older than the config/environment setting HB_DELETE_AFTER

Note that the use of sqlalchemy allows us to switch to a different DB in the future (e.g., Postgres) if we decide to do that, with only a trivial change to the sqlalchemy engine object.

[sybenzvi]

It looks like we need a unit test for this.

[KaraMelih]

I think, I am happy with this version. Database works as expected, I added a temporary fix for the factorial problem mentioned here #100

I updated the poetry dependencies to include sqlalchemy and tabulate

tabulate is used when someone requests a

snews_pt write-hb-logs

then it dumps the table on the console (on the server)

[KaraMelih]

One thing that might not be desirable is the precision of the Latencies.
Here;

SNEWS_Coincidence_System/snews_cs/snews_hb.py

Line 129 in 486f316

msg["Latency"] = msg["Latency"].astype('timedelta64[s]').astype(int)

I converted them to integer seconds because sqlalchemy did not like the timedelta objects. However, we are loosing some precision. We should decide if want to improve that