properly close the connections on drop/garbage collection

properly close the connections on drop/garbage collection

…es not worth own commit)

Merge branch 'postgres-DB-handler' of github.com:einhundert/pytest-monitor into postgres-DB-handler

with-context: Remove the with-contexts inside the PostgresDBHandler and
do manual commit after cursor execution instead to prevent closing of
the database connection.

db-close: Add close() function to both db-handlers and the
PytestMonitorSession class in order to properly close all database
connections at the end of the tests. This happens in pytest-monitor.py
with the pytest_sessionfinish hook implementation that calls the
close() function on the PytestMonitorSession object.

env-id: The query for the env-id PostgresDBHandler.get_env_id() queries
for the env id in the database and gets a tuple from the query but does
return it without unwrapping the value inside. This leads to problems
as the value is meant to be unwrapped, therefore return the unwrapped
value.

This is a helper function for generating a session description in
session.py.

after tests are done.

The session is tested for proper closing of the database connection.
The close() function is called from the hookimplementation in
pytest_sessionfinish().

The dbhandlers are tested for proper setup of the tables and in
consequence a working connection. The PostgresDBHandler requires a
running postgres server, that's what the docker-compose.yml file is for

Merge with upstream master and update changelog.rst

…nitor into postgres-DB-handler

during testing.

…nitor into postgres-DB-handler

The memory profiler only reports Exceptions when it should report all
Exceptions that inherit from BaseExceptions.

This is ultimately reworked in a PR incorporating a simplified
memory profiler module into the codebase that fixes not only this issue
but also gives the possibility of getting measurements for failing
tests.

This workaround uses return values to work around the issue. This way
pytest-monitor can check for BaseExceptions and act accordingly.
Like described earlier the ultimate goal should probably be to replace
the whole memory profiler as proposed in this PR:
CFMTech#82