Fix load when PVs missing

[mdoucet]

It happens too often that during changing the configuration of the instrument, PV related to the state of the analyzer or polarizer are missing. In this case the load fails because of the meta data is incomplete. In 95% of the cases it's a case where we are not measuring a polarized state and the device is removed and the PVs not set properly.

This PR does a number of things:

• It falls back to another method of parsing the events which work in nearly all cases, but is just slower.
• It also makes the way we deal with the dead time correction uniform (it was previously ignored for this second filtering type).

[backmari]

This PR changes the loading to fall back to "simple filtering" if the PV:s for the polarizer and/or analyzer state is missing. It also makes the dead time correction apply regardless of the filtering method.
(Since Codecov doesn't seem to work quite right, I can't easily see what the test coverage is like, but if needed we can add a test later.)