Enable reusing past labels DATANG-3799

[samsucik]

Jira task

How I designed this:

• one can add past labels (where available) by switching a toggle
• the labels are searched for across all the files in the sidebar, based on matching both the generated text column as well as all of the columns available to be used as inputs
• these past labels are stored in a separate column but otherwise get treated like human labels in computing the overall score
• when a fresh human label is available as well, it takes precedence over the past label
• where multiple (contradicting) labels are available for a row, the most recent one is used (relying on the time-based ordering of datafiles in the sidebar)
• in addition, one can choose to ignore (skip) rows with past labels to speed up the manual labelling process

All of these design decisions can be challenged but the current design at least supports @vikion in her efforts 🙂

[sweco]

@samsucik, finally got to this 😊 Looks good, I only marked a few places that I found a bit harder to understand.