PandasReporter using daily sensor as input misses value on spring DST transition

[Flix6x]

Two underlying issues:

1. search_data_sources and search_beliefs use slightly different filters for event timing
2. Passing a one-day resolution to search_beliefs leads to clipping off the 23-hour day of the spring DST transition.

Issue 1

• search_beliefs uses event_ends_after and event_starts_before internally, resulting in events that partly fall in the searched time window to be included. After resampling, the results are clipped to the desired time interval.
• search_data_sources uses event_starts_after and event_ends_before, resulting in event that partly fall in the searched time window to be excluded. On the day of the spring DST transition, the 23-hour event is being interpreted as a 24-hour event that extends 1 one hour past midnight.

Related to #484.

Issue 2

Given some sensor (here, with ID 40755) with a "P1D" event resolution in the "Europe/Amsterdam" timezone, and an event registered on that sensor with event start at "2023-03-26 00:00:00+01:00" (day of spring DST):

from flexmeasures import Sensor
import pandas as pd
sensor = Sensor.query.get(40755)

# one row, as expected
sensor.search_beliefs(event_starts_after="2023-03-26 00:00:00+01:00", event_ends_before="2023-03-27 00:00:00+02:00")

# empty frame, unexpected
sensor.search_beliefs(event_starts_after="2023-03-26 00:00:00+01:00", event_ends_before="2023-03-27 00:00:00+02:00", resolution=pd.Timedelta("P1D"))