Add pre_ping.

Indeed, before this patch, we weren't retrying to reconnect properly.
This patch fixes the issue by using the pre_ping function of sqlalchemy.

This patch fixes #346.

Contributors:
  * @spirillen

PyFunceble/database/session.py

@@ -150,6 +150,7 @@ def init_db_sessions(self) -> "DBSession":
                 autoflush=True,
                 autocommit=False,
                 expire_on_commit=False,
+                pre_ping=True,
             )
 
         return self
@@ -180,7 +181,11 @@ def get_new_session(self) -> sqlalchemy.orm.sessionmaker:
         )
 
         return sqlalchemy.orm.sessionmaker(
-            bind=engine, autoflush=True, autocommit=False, expire_on_commit=False
+            bind=engine,
+            autoflush=True,
+            autocommit=False,
+            expire_on_commit=False,
+            pre_ping=True,
         )
 
     @execute_if_authorized(None)
@@ -195,7 +200,11 @@ def get_new_pool_session(self) -> sqlalchemy.orm.sessionmaker:
         )
 
         return sqlalchemy.orm.sessionmaker(
-            bind=engine, autoflush=True, autocommit=False, expire_on_commit=False
+            bind=engine,
+            autoflush=True,
+            autocommit=False,
+            expire_on_commit=False,
+            pre_ping=True,
         )
 
     def close(self) -> "DBSession":