As Fluentd tail plugin can send data from a single file several packets,
this needs to be handled properly on ingestion side.

As timestamps are always different, checking for a null rate is pointless.

Items generated during the creation of a Data Analysis are now given unique names and references even with multiple outputs

See merge request !165

After loading the data from the ZODB, the 'getBucketBy...' methods
now free unused memory. Before this patch, it was the responsibility
of the user to deactivate no longer needed buckets, now this happens
automatically.

Before this patch 'getBucketByIndex' was buggy & would fail with

    AttributeError: 'str' object has no attribute 'value'

But it wasn't recognized, because this method wasn't tested before. This
patch adds a test to the method & fixes it.

See merge request nexedi/wendelin!163

See merge request !162

See merge request nexedi/wendelin!161

This reverts commit 111919f5

As noted by @jm in nexedi/wendelin!160 (comment 199327)
'self.tic() automatically commits first'. So let's drop redundant code
in tests.

Needed after 85b18536
(uses old XML format which breaks codign styles tests)

See merge request !158

levin: This cherry-picks the latest version of the
'DataAnalysis_clearAndReprocessFromScratch' script from a customer
project. It's needed to successfully run the 'refresh' workflow [1]
and was perhaps forgotten to be added in nexedi/wendelin!31.

[1] https://lab.nexedi.com/nexedi/wendelin/blob/04787344/bt5/erp5_wendelin/SkinTemplateItem/portal_skins/erp5_wendelin/ERP5Site_executeDataAnalysisList.py#L37

Fix.

context) have this format.

sensor.product

Thus be consistent also here for Data Streams' reference.

(https://wendelin.nexedi.com/wendelin-HowTo.Create.Notebook.with.Plotly.Chart.Editor/wendelin-Poltly.Chart.Editor.4?format=)